Issue 18709: The SysML classification of properties is incomplete (sysml-rtf) Source: NASA (Dr. Nicolas F. Rouquette, nicolas.f.rouquette(at)jpl.nasa.gov) Nature: Uncategorized Issue Severity: Summary: SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. … A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload – I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems — what would it mean to send / receive such signals? Resolution: Defer Postponed to the next RTF Revised Text: Actions taken: May 13, 2013: received issue January 3, 2017: Deferred April 6, 2017: closed issue Discussion: End of Annotations:===== m: "Rouquette, Nicolas F (313K)" To: "issues@omg.org" CC: "sysml-rtf@omg.org" Subject: The SysML classification of properties is incomplete Thread-Topic: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EA== Date: Mon, 13 May 2013 18:11:21 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: user-agent: Microsoft-MacOutlook/14.3.4.130416 x-originating-ip: [128.149.137.114] X-Source-Sender: nicolas.f.rouquette@jpl.nasa.gov X-AUTH: Authorized X-Virus-Scanned: amavisd-new at omg.org X-Brightmail-Tracker: AAAAAR16Wag= X-Brightmail-Tracker: AAAAAA== SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=from:to:cc:references:in-reply-to:subject:date:message-id :mime-version:content-type:x-mailer:thread-index:content-language; bh=RekYijL83JpDTx5iHYEt1SfDfeDsA4t+5nvzk5/SiBY=; b=N8CQEFF5n0kW3oH4O6pAAjmi/i6RIp68x9OSxb5Z8iAfsoEEb/euHpKqnu+cGWC6yZ o78Aczu3Ai4h7Cur5jcIIbHZgZ/Xoo2oauf8A8c+vLiA0WHBjXJgK8SbnA0VCXd8SHq0 OkT+7svyCK+VwExkCX86lYZWiMbqdHP858Z0Zvm/OsaCB9Gt4r0HIwnJqzouUTZLfwwN MyiPecSKcfFNUHc7GJR+T0nNLcc/K25aHDDZHNNuefDD41c26vOzaStrVSBlDVGk0R24 1X6wWS+PboY9KlqJIBEl5RNt2qYUYHsZKgie5d2gsaVPXwkFPX3P8bPyAsuB1gw17dRH i/cQ== X-Received: by 10.229.148.136 with SMTP id p8mr2656051qcv.15.1369659929817; Mon, 27 May 2013 06:05:29 -0700 (PDT) From: "Sanford Friedenthal" To: "'Rouquette, Nicolas F \(313K\)'" Cc: Subject: RE: The SysML classification of properties is incomplete Date: Mon, 27 May 2013 09:05:26 -0400 X-Mailer: Microsoft Office Outlook 12.0 Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1g X-Virus-Scanned: amavisd-new at omg.org Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Chonoles, Michael J" To: "safriedenthal@gmail.com" , "'Rouquette, Nicolas F (313K)'" CC: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYA= Date: Mon, 27 May 2013 17:13:11 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [158.186.156.88] X-Proofpoint-Virus-Version: vendor=fsecure engine=2.50.10432:5.10.8626,1.0.431,0.0.0000 definitions=2013-05-27_04:2013-05-27,2013-05-27,1970-01-01 signatures=0 X-Virus-Scanned: amavisd-new at omg.org Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Rouquette, Nicolas F (313K)" To: Sanford Friedenthal CC: "sysml-rtf@omg.org" , "Karban, Robert (3101-Affiliate)" , "Jenkins, J Steven (3101)" Subject: Re: The SysML classification of properties is incomplete Thread-Topic: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggACG6oA= Date: Mon, 27 May 2013 21:03:41 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: user-agent: Microsoft-MacOutlook/14.3.4.130416 x-originating-ip: [128.149.137.113] X-Source-Sender: nicolas.f.rouquette@jpl.nasa.gov X-AUTH: Authorized X-Virus-Scanned: amavisd-new at omg.org Sandy, You seem to be using blocks for modeling structured data; this is a notorious source of confusion and errors in practice. Below, I've summarized 5 guidelines on this topic (Robert Karban and I have been working on developing guidelines for a restricted subset of SysML at JPL): 1) distinguish between two kinds of blocks: Blocks that model the structure of the system Blocks that model the structure of runtime objects (not runtime data) A car is an example of the former . at runtime, it just exists. A car repair bill is example of the latter if is modeled with identity (I.e., a runtime object that can be referenced but not copied) Blocks intended for modeling runtime objects must provide an identity criteria with either or both of the following: - at least one value property attribute with {isID = true} whose type maps directly or transitively to xs:ID see: http://www.w3.org/TR/2012/REC-xmlschema11-1-20120405/structures.html#sec-cvc-simple-type - a single value property attribute with {isID = true} whose type maps directly or transitively to xs:ENTITY see: http://www.w3.org/TR/2012/REC-xmlschema11-1-20120405/structures.html#sec-cvc-id 2) SysML ValueTypes model runtime data . primitive or tree-structured. Primitive SysML ValueTypes have no value properties. Tree-structured SysML ValueTypes have at least one value property. In either case, a SysML ValueType models runtime data. Fundamentally, runtime data has only a value that can be copied but no identity that can be referenced in any way. If anything, its value is its identity. There cannot be a value property attribute with {isID=true} or whose type maps directly or transitively to xs:ID or xs:ENTITY In either case, a SysML ValueType may have a unit and may have a quantityKind. A tree-structured SysML ValueType may have a unit (respectively quantityKind) if and only if all of its value property attributes are typed by SysML ValueTypes with no unit (respectively, no quantityKind). 3) SysML's PrimitiveValueTypes specify no mapping in the spec. This is intended to give SysML users and tools the flexibility to map them by specialization to one or more atomic datatypes. Such mapping is done via the "schemaType" MOF/XMI tag (see Table 7.1 in MOF/XMI 2.4.1) and "anyAtomicType" -- see: http://www.w3.org/TR/2012/REC-xmlschema11-2-20120405/#anyAtomicType 4) structured SysML ValueTypes can be mapped to a complex type in XML Schema 1.1 This is a variant of the above; the difference is that the "schemaType" MOF/XMI tag points the the URI of an XSD complex type . see: http://www.w3.org/TR/2012/REC-xmlschema11-1-20120405/structures.html#Complex_Type_Definitions In this case, there must be a 1-1 correspondence between the value property attributes of the structured SysML ValueType and the attributes of the XSD complex type. 5) Any structured SysML ValueType must be constrained to represent tree structured data only. The representation of a runtime structured data is an InstanceSpecification classified by a structured SysML ValueType. If a structured SysML ValueType is used for typing one of its value property attributes, then constraints must be explicitly added to prevent representing circular structured data since such data could not be serialized properly as a well-formed instance of an XSD complex type. - Nicolas. From: Sanford Friedenthal Date: Monday, May 27, 2013 6:05 AM To: JPL Cc: "sysml-rtf@omg.org" Subject: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Rouquette, Nicolas F (313K)" To: "Chonoles, Michael J" , "safriedenthal@gmail.com" CC: "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAA== Date: Mon, 27 May 2013 21:05:26 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: user-agent: Microsoft-MacOutlook/14.3.4.130416 x-originating-ip: [128.149.137.113] X-Source-Sender: nicolas.f.rouquette@jpl.nasa.gov X-AUTH: Authorized X-Virus-Scanned: amavisd-new at omg.org Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Chonoles, Michael J" To: "Rouquette, Nicolas F (313K)" , "safriedenthal@gmail.com" CC: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYw Date: Mon, 27 May 2013 21:11:34 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [158.186.156.88] X-Proofpoint-Virus-Version: vendor=fsecure engine=2.50.10432:5.10.8626,1.0.431,0.0.0000 definitions=2013-05-27_05:2013-05-27,2013-05-27,1970-01-01 signatures=0 X-Virus-Scanned: amavisd-new at omg.org If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Rouquette, Nicolas F (313K)" To: "Chonoles, Michael J" , "safriedenthal@gmail.com" CC: "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwA= Date: Mon, 27 May 2013 23:31:45 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: user-agent: Microsoft-MacOutlook/14.3.4.130416 x-originating-ip: [128.149.137.113] X-Source-Sender: nicolas.f.rouquette@jpl.nasa.gov X-AUTH: Authorized X-Virus-Scanned: amavisd-new at omg.org Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Chonoles, Michael J" To: "Rouquette, Nicolas F (313K)" , "safriedenthal@gmail.com" CC: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwCAAABSEA== Date: Mon, 27 May 2013 23:34:30 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [158.186.156.88] X-Proofpoint-Virus-Version: vendor=fsecure engine=2.50.10432:5.10.8626,1.0.431,0.0.0000 definitions=2013-05-27_05:2013-05-27,2013-05-27,1970-01-01 signatures=0 X-Virus-Scanned: amavisd-new at omg.org I can throw a brick at you, that would be an asynch signal that I.m mad at you. «grin» I don.t see why I must use an operation to do so. So, I'd recommend that signal properties be allowed to be blocks. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 7:32 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Rouquette, Nicolas F (313K)" To: "Chonoles, Michael J" , "safriedenthal@gmail.com" CC: "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwCAAABSEIAADZMA Date: Tue, 28 May 2013 00:21:28 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: user-agent: Microsoft-MacOutlook/14.3.4.130416 x-originating-ip: [128.149.137.113] X-Source-Sender: nicolas.f.rouquette@jpl.nasa.gov X-AUTH: Authorized X-Virus-Scanned: amavisd-new at omg.org Michael, Again, I said that, as far as the SysML spec is concerned, it is unclear whether we can throw bricks or not. I agree with you that we should be able to play such games; however, there are potential dangers involved when a block is used for two purposes in a model: modeling some structural aspect of a system itself (e.g., a brick in a model for analyzing the structure of a house) modeling something that is involved in exchanges that are modeled via operation parameters or signal properties and possibly other constructs (e.g., a brick in a model for building a house) The meaning of the model can become dicey when a block is used for both purposes: structure and behavior . e.g., analyzing the structural properties of brick houses and analyzing supply chain operations for building houses. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 4:34 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete I can throw a brick at you, that would be an asynch signal that I.m mad at you. «grin» I don.t see why I must use an operation to do so. So, I'd recommend that signal properties be allowed to be blocks. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 7:32 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Chonoles, Michael J" To: "Rouquette, Nicolas F (313K)" , "safriedenthal@gmail.com" CC: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwCAAABSEIAADZMAgAAjN2A= Date: Tue, 28 May 2013 02:28:56 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [158.186.156.88] X-Proofpoint-Virus-Version: vendor=fsecure engine=2.50.10432:5.10.8626,1.0.431,0.0.0000 definitions=2013-05-28_01:2013-05-27,2013-05-28,1970-01-01 signatures=0 X-Virus-Scanned: amavisd-new at omg.org I.m not sure why you see that as problematic. Certainly when we have flows, we can have ammunition (or bricks) flow. They are both important parts of systems modeling. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 8:21 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Again, I said that, as far as the SysML spec is concerned, it is unclear whether we can throw bricks or not. I agree with you that we should be able to play such games; however, there are potential dangers involved when a block is used for two purposes in a model: modeling some structural aspect of a system itself (e.g., a brick in a model for analyzing the structure of a house) modeling something that is involved in exchanges that are modeled via operation parameters or signal properties and possibly other constructs (e.g., a brick in a model for building a house) The meaning of the model can become dicey when a block is used for both purposes: structure and behavior . e.g., analyzing the structural properties of brick houses and analyzing supply chain operations for building houses. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 4:34 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete I can throw a brick at you, that would be an asynch signal that I.m mad at you. «grin» I don.t see why I must use an operation to do so. So, I'd recommend that signal properties be allowed to be blocks. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 7:32 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Rouquette, Nicolas F (313K)" To: "Chonoles, Michael J" , "safriedenthal@gmail.com" CC: "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwCAAABSEIAADZMAgAAjN2CAAAYRgA== Date: Tue, 28 May 2013 02:49:13 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: user-agent: Microsoft-MacOutlook/14.3.4.130416 x-originating-ip: [128.149.137.113] X-Source-Sender: nicolas.f.rouquette@jpl.nasa.gov X-AUTH: Authorized X-Virus-Scanned: amavisd-new at omg.org Michael, The importance of distinguishing between: 1 a block modeling a structured data value 2 a block modeling a data object Is that some operations are legal with one but not the other. 1: can be copied; can be tested for equality 2: cannot be copied (without creating a new object with a distinct identity than the original); can be queried for its identity (if there is an identifying value property) Isn't this important enough to you? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 7:28 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete I.m not sure why you see that as problematic. Certainly when we have flows, we can have ammunition (or bricks) flow. They are both important parts of systems modeling. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 8:21 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Again, I said that, as far as the SysML spec is concerned, it is unclear whether we can throw bricks or not. I agree with you that we should be able to play such games; however, there are potential dangers involved when a block is used for two purposes in a model: modeling some structural aspect of a system itself (e.g., a brick in a model for analyzing the structure of a house) modeling something that is involved in exchanges that are modeled via operation parameters or signal properties and possibly other constructs (e.g., a brick in a model for building a house) The meaning of the model can become dicey when a block is used for both purposes: structure and behavior . e.g., analyzing the structural properties of brick houses and analyzing supply chain operations for building houses. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 4:34 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete I can throw a brick at you, that would be an asynch signal that I.m mad at you. «grin» I don.t see why I must use an operation to do so. So, I'd recommend that signal properties be allowed to be blocks. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 7:32 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: Matthew Hause To: "Chonoles, Michael J" , "Rouquette, Nicolas F (313K)" , "safriedenthal@gmail.com" CC: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwCAAABSEIAATY3w Date: Tue, 28 May 2013 04:11:28 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [108.253.244.82] x-esetresult: clean, is OK x-esetid: 350DFF23A501E0776E4BA4 X-Virus-Scanned: amavisd-new at omg.org Also, interactions can also consist of physical entities as well. I agree with Michael. We should bring a brick to Berlin to demonstrate. J From: Chonoles, Michael J [mailto:michael.j.chonoles@lmco.com] Sent: Monday, May 27, 2013 6:35 PM To: Rouquette, Nicolas F (313K); safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete I can throw a brick at you, that would be an asynch signal that I.m mad at you. «grin» I don.t see why I must use an operation to do so. So, I'd recommend that signal properties be allowed to be blocks. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 7:32 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: Matthew Hause To: "Rouquette, Nicolas F (313K)" , "Chonoles, Michael J" , "safriedenthal@gmail.com" CC: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwCAAABSEIAADZMAgABAbtA= Date: Tue, 28 May 2013 04:15:01 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [108.253.244.82] x-esetresult: clean, is OK x-esetid: 350DFF23A501E0776E4BA4 X-Virus-Scanned: amavisd-new at omg.org I disagree. They can and should be used for both purposes. Otherwise, SysML risks becoming a language just for modeling software intensive systems and not all types of systems. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 7:21 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Again, I said that, as far as the SysML spec is concerned, it is unclear whether we can throw bricks or not. I agree with you that we should be able to play such games; however, there are potential dangers involved when a block is used for two purposes in a model: modeling some structural aspect of a system itself (e.g., a brick in a model for analyzing the structure of a house) modeling something that is involved in exchanges that are modeled via operation parameters or signal properties and possibly other constructs (e.g., a brick in a model for building a house) The meaning of the model can become dicey when a block is used for both purposes: structure and behavior . e.g., analyzing the structural properties of brick houses and analyzing supply chain operations for building houses. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 4:34 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete I can throw a brick at you, that would be an asynch signal that I.m mad at you. «grin» I don.t see why I must use an operation to do so. So, I'd recommend that signal properties be allowed to be blocks. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 7:32 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Rouquette, Nicolas F (313K)" To: Matthew Hause , "Chonoles, Michael J" , "safriedenthal@gmail.com" CC: "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwCAAABSEIAADZMAgABAbtCAAANyAA== Date: Tue, 28 May 2013 04:24:23 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: user-agent: Microsoft-MacOutlook/14.3.4.130416 x-originating-ip: [128.149.137.113] X-Source-Sender: nicolas.f.rouquette@jpl.nasa.gov X-AUTH: Authorized X-Virus-Scanned: amavisd-new at omg.org So you're OK in modeling in SysML a car with wheels where car and wheel are blocks and allowing the possibility for the car to throw a wheel at something. Well, I guess you're more advanced SysML modelers than we are at JPL. If that's what you're doing and want to do, please enlighten us and show us how you do this without getting into Russell's paradox (I.e., self-referentiality) - Nicolas. From: Matthew Hause Date: Monday, May 27, 2013 9:15 PM To: JPL , Michael Chonoles , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete I disagree. They can and should be used for both purposes. Otherwise, SysML risks becoming a language just for modeling software intensive systems and not all types of systems. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 7:21 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Again, I said that, as far as the SysML spec is concerned, it is unclear whether we can throw bricks or not. I agree with you that we should be able to play such games; however, there are potential dangers involved when a block is used for two purposes in a model: modeling some structural aspect of a system itself (e.g., a brick in a model for analyzing the structure of a house) modeling something that is involved in exchanges that are modeled via operation parameters or signal properties and possibly other constructs (e.g., a brick in a model for building a house) The meaning of the model can become dicey when a block is used for both purposes: structure and behavior . e.g., analyzing the structural properties of brick houses and analyzing supply chain operations for building houses. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 4:34 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete I can throw a brick at you, that would be an asynch signal that I.m mad at you. «grin» I don.t see why I must use an operation to do so. So, I'd recommend that signal properties be allowed to be blocks. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 7:32 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. X-Trusted-NM: yes Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete From: Nerijus Jankevicius Date: Tue, 28 May 2013 16:26:03 +0300 Cc: "Chonoles, Michael J" , "safriedenthal@gmail.com" , "sysml-rtf@omg.org" To: "Rouquette, Nicolas F (313K)" X-Mailer: Apple Mail (2.1085) X-Virus-Scanned: amavisd-new at omg.org Does SysML allow typing operation parameters, behavior parameters and pins by UML primitive types? Can we use Enumerations which are not stereotyped by ValueType (like VerdictKind is) ? If no, where it is described and constrained? Thanks. Nerijus On May 28, 2013, at 2:31 AM, Rouquette, Nicolas F (313K) wrote: Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Rouquette, Nicolas F (313K)" To: Nerijus Jankevicius CC: "Chonoles, Michael J" , "safriedenthal@gmail.com" , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwCAAV52gP//kM+A Date: Tue, 28 May 2013 13:48:09 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: user-agent: Microsoft-MacOutlook/14.3.4.130416 x-originating-ip: [128.149.137.113] X-Source-Sender: nicolas.f.rouquette@jpl.nasa.gov X-AUTH: Authorized X-Virus-Scanned: amavisd-new at omg.org Yes; ValueType can be applied to a PrimitiveType and to an Enumeration; both are kinds of DataType. You already filed an issue about VerdictKind missing the ValueType stereotype -- http://www.omg.org/issues/sysml-rtf.html#Issue18312 Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 6:26 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Does SysML allow typing operation parameters, behavior parameters and pins by UML primitive types? Can we use Enumerations which are not stereotyped by ValueType (like VerdictKind is) ? If no, where it is described and constrained? Thanks. Nerijus On May 28, 2013, at 2:31 AM, Rouquette, Nicolas F (313K) wrote: Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. X-Trusted-NM: yes Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete From: Nerijus Jankevicius Date: Tue, 28 May 2013 17:07:04 +0300 Cc: "Chonoles, Michael J" , "safriedenthal@gmail.com" , "sysml-rtf@omg.org" To: "Rouquette, Nicolas F (313K)" X-Mailer: Apple Mail (2.1085) X-Virus-Scanned: amavisd-new at omg.org Nicolas, You misunderstood my question. I'm asking about usage of UML primitive types in SysML models. Is this restricted or not? If yes, where in SysML spec? I used VerdictKind as example. Nerijus On May 28, 2013, at 4:48 PM, Rouquette, Nicolas F (313K) wrote: Yes; ValueType can be applied to a PrimitiveType and to an Enumeration; both are kinds of DataType. You already filed an issue about VerdictKind missing the ValueType stereotype -- http://www.omg.org/issues/sysml-rtf.html#Issue18312 Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 6:26 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Does SysML allow typing operation parameters, behavior parameters and pins by UML primitive types? Can we use Enumerations which are not stereotyped by ValueType (like VerdictKind is) ? If no, where it is described and constrained? Thanks. Nerijus On May 28, 2013, at 2:31 AM, Rouquette, Nicolas F (313K) wrote: Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Rouquette, Nicolas F (313K)" To: Nerijus Jankevicius CC: "Chonoles, Michael J" , "safriedenthal@gmail.com" , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwCAAV52gP//kM+AgAB6pgD//455AA== Date: Tue, 28 May 2013 14:20:47 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: user-agent: Microsoft-MacOutlook/14.3.4.130416 x-originating-ip: [128.149.137.113] X-Source-Sender: nicolas.f.rouquette@jpl.nasa.gov X-AUTH: Authorized X-Virus-Scanned: amavisd-new at omg.org Nerijus, I understood your question; the answer is still yes. SysML ValueType extends DataType. This means ValueType can be applied to any kind of UML DataType, including specializations; I.e., PrimitiveType and Enumeration. Note that a value property in SysML must be typed by a SysML ValueType, even if it is a PrimitiveType or Enumeration. A property typed by a non-ValueType PrimitiveType or Enumeration is not, by definition, a value property in SysML. Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 7:07 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas, You misunderstood my question. I'm asking about usage of UML primitive types in SysML models. Is this restricted or not? If yes, where in SysML spec? I used VerdictKind as example. Nerijus On May 28, 2013, at 4:48 PM, Rouquette, Nicolas F (313K) wrote: Yes; ValueType can be applied to a PrimitiveType and to an Enumeration; both are kinds of DataType. You already filed an issue about VerdictKind missing the ValueType stereotype -- http://www.omg.org/issues/sysml-rtf.html#Issue18312 Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 6:26 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Does SysML allow typing operation parameters, behavior parameters and pins by UML primitive types? Can we use Enumerations which are not stereotyped by ValueType (like VerdictKind is) ? If no, where it is described and constrained? Thanks. Nerijus On May 28, 2013, at 2:31 AM, Rouquette, Nicolas F (313K) wrote: Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. X-Trusted-NM: yes Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete From: Nerijus Jankevicius Date: Tue, 28 May 2013 17:34:59 +0300 Cc: "Chonoles, Michael J" , "safriedenthal@gmail.com" , "sysml-rtf@omg.org" To: "Rouquette, Nicolas F (313K)" X-Mailer: Apple Mail (2.1085) X-Virus-Scanned: amavisd-new at omg.org Nicolas, I can't see how your answer is related to my question. I'm not asking if ValueType can be applied to datatypes or primitives, it is clear. I'm not asking about value properties too. I'm asking about Operation parameters, behavior parameters and pin types. They are not value properties. Nerijus On May 28, 2013, at 5:20 PM, Rouquette, Nicolas F (313K) wrote: Nerijus, I understood your question; the answer is still yes. SysML ValueType extends DataType. This means ValueType can be applied to any kind of UML DataType, including specializations; I.e., PrimitiveType and Enumeration. Note that a value property in SysML must be typed by a SysML ValueType, even if it is a PrimitiveType or Enumeration. A property typed by a non-ValueType PrimitiveType or Enumeration is not, by definition, a value property in SysML. Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 7:07 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas, You misunderstood my question. I'm asking about usage of UML primitive types in SysML models. Is this restricted or not? If yes, where in SysML spec? I used VerdictKind as example. Nerijus On May 28, 2013, at 4:48 PM, Rouquette, Nicolas F (313K) wrote: Yes; ValueType can be applied to a PrimitiveType and to an Enumeration; both are kinds of DataType. You already filed an issue about VerdictKind missing the ValueType stereotype -- http://www.omg.org/issues/sysml-rtf.html#Issue18312 Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 6:26 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Does SysML allow typing operation parameters, behavior parameters and pins by UML primitive types? Can we use Enumerations which are not stereotyped by ValueType (like VerdictKind is) ? If no, where it is described and constrained? Thanks. Nerijus On May 28, 2013, at 2:31 AM, Rouquette, Nicolas F (313K) wrote: Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Rouquette, Nicolas F (313K)" To: Nerijus Jankevicius CC: "Chonoles, Michael J" , "safriedenthal@gmail.com" , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwCAAV52gP//kM+AgAB6pgD//455AIAAeVSA//+M5QA= Date: Tue, 28 May 2013 14:43:03 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: user-agent: Microsoft-MacOutlook/14.3.4.130416 x-originating-ip: [128.149.137.113] X-Source-Sender: nicolas.f.rouquette@jpl.nasa.gov X-AUTH: Authorized X-Virus-Scanned: amavisd-new at omg.org Nerijus, Answer is still yes. See SysML 1.3, section 8.3.1.1.1: 8.3.1.1.1 Block and ValueType Definitions A SysML Block defines a collection of features to describe a system or other element of interest. A SysML ValueType defines values that may be used within a model. SysML blocks are based on UML classes, as extended by UML composite structures. SysML value types are based on UML data types. I admit that this paragraph could be clearer to avoid so many contentious debates. In your example: Typing an operation or behavior parameter or activity pin by a UML DataType of any kind is OK as long as this UML DataType has the SysML ValueType stereotype applied. Without applying SysML ValueType to any kind of UML DataType, the meaning of the UML+SysML model falls outside the scope of the SysML spec. UML says something about this but SysML doesn't. - Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 7:34 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas, I can't see how your answer is related to my question. I'm not asking if ValueType can be applied to datatypes or primitives, it is clear. I'm not asking about value properties too. I'm asking about Operation parameters, behavior parameters and pin types. They are not value properties. Nerijus On May 28, 2013, at 5:20 PM, Rouquette, Nicolas F (313K) wrote: Nerijus, I understood your question; the answer is still yes. SysML ValueType extends DataType. This means ValueType can be applied to any kind of UML DataType, including specializations; I.e., PrimitiveType and Enumeration. Note that a value property in SysML must be typed by a SysML ValueType, even if it is a PrimitiveType or Enumeration. A property typed by a non-ValueType PrimitiveType or Enumeration is not, by definition, a value property in SysML. Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 7:07 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas, You misunderstood my question. I'm asking about usage of UML primitive types in SysML models. Is this restricted or not? If yes, where in SysML spec? I used VerdictKind as example. Nerijus On May 28, 2013, at 4:48 PM, Rouquette, Nicolas F (313K) wrote: Yes; ValueType can be applied to a PrimitiveType and to an Enumeration; both are kinds of DataType. You already filed an issue about VerdictKind missing the ValueType stereotype -- http://www.omg.org/issues/sysml-rtf.html#Issue18312 Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 6:26 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Does SysML allow typing operation parameters, behavior parameters and pins by UML primitive types? Can we use Enumerations which are not stereotyped by ValueType (like VerdictKind is) ? If no, where it is described and constrained? Thanks. Nerijus On May 28, 2013, at 2:31 AM, Rouquette, Nicolas F (313K) wrote: Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Chonoles, Michael J" To: "Rouquette, Nicolas F (313K)" , "safriedenthal@gmail.com" CC: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwCAAABSEIAADZMAgAAjN2CAAAYRgIABGz4w Date: Tue, 28 May 2013 19:44:15 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [158.186.156.88] X-Proofpoint-Virus-Version: vendor=fsecure engine=2.50.10432:5.10.8626,1.0.431,0.0.0000 definitions=2013-05-28_08:2013-05-28,2013-05-28,1970-01-01 signatures=0 X-Virus-Scanned: amavisd-new at omg.org Yes, I.m just saying the a signal could be used to move either type of block. Isn.t throwing a brick a signal? From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 10:49 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, The importance of distinguishing between: 1 a block modeling a structured data value 2 a block modeling a data object Is that some operations are legal with one but not the other. 1: can be copied; can be tested for equality 2: cannot be copied (without creating a new object with a distinct identity than the original); can be queried for its identity (if there is an identifying value property) Isn't this important enough to you? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 7:28 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete I.m not sure why you see that as problematic. Certainly when we have flows, we can have ammunition (or bricks) flow. They are both important parts of systems modeling. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 8:21 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Again, I said that, as far as the SysML spec is concerned, it is unclear whether we can throw bricks or not. I agree with you that we should be able to play such games; however, there are potential dangers involved when a block is used for two purposes in a model: modeling some structural aspect of a system itself (e.g., a brick in a model for analyzing the structure of a house) modeling something that is involved in exchanges that are modeled via operation parameters or signal properties and possibly other constructs (e.g., a brick in a model for building a house) The meaning of the model can become dicey when a block is used for both purposes: structure and behavior . e.g., analyzing the structural properties of brick houses and analyzing supply chain operations for building houses. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 4:34 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete I can throw a brick at you, that would be an asynch signal that I.m mad at you. «grin» I don.t see why I must use an operation to do so. So, I'd recommend that signal properties be allowed to be blocks. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 7:32 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Chonoles, Michael J" To: "Rouquette, Nicolas F (313K)" , "Matthew Hause" , "safriedenthal@gmail.com" CC: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwCAAABSEIAADZMAgABAbtCAAANyAIABAfKw Date: Tue, 28 May 2013 19:54:14 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [158.186.156.88] X-Proofpoint-Virus-Version: vendor=fsecure engine=2.50.10432:5.10.8626,1.0.431,0.0.0000 definitions=2013-05-28_08:2013-05-28,2013-05-28,1970-01-01 signatures=0 X-Virus-Scanned: amavisd-new at omg.org My ship contains launchers and missiles. I want to launch the missiles and hit an incoming target. I call an operation on a launcher which sends a signal containing a missile to the target. I don.t think of this as an operation on the incoming target If I was modeling this, and I often model both sides, the activity diagram for my ship would have a swimlane for the launcher which would have a send event within it, passing the missile along. (It would also decrement the count of missiles I have) The activity diagram for the incoming target would have a wait for event (and hopefully, destruct «grin») Michael From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Tuesday, May 28, 2013 12:24 AM To: Matthew Hause; Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete So you're OK in modeling in SysML a car with wheels where car and wheel are blocks and allowing the possibility for the car to throw a wheel at something. Well, I guess you're more advanced SysML modelers than we are at JPL. If that's what you're doing and want to do, please enlighten us and show us how you do this without getting into Russell's paradox (I.e., self-referentiality) - Nicolas. From: Matthew Hause Date: Monday, May 27, 2013 9:15 PM To: JPL , Michael Chonoles , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete I disagree. They can and should be used for both purposes. Otherwise, SysML risks becoming a language just for modeling software intensive systems and not all types of systems. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 7:21 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Again, I said that, as far as the SysML spec is concerned, it is unclear whether we can throw bricks or not. I agree with you that we should be able to play such games; however, there are potential dangers involved when a block is used for two purposes in a model: modeling some structural aspect of a system itself (e.g., a brick in a model for analyzing the structure of a house) modeling something that is involved in exchanges that are modeled via operation parameters or signal properties and possibly other constructs (e.g., a brick in a model for building a house) The meaning of the model can become dicey when a block is used for both purposes: structure and behavior . e.g., analyzing the structural properties of brick houses and analyzing supply chain operations for building houses. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 4:34 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete I can throw a brick at you, that would be an asynch signal that I.m mad at you. «grin» I don.t see why I must use an operation to do so. So, I'd recommend that signal properties be allowed to be blocks. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 7:32 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Chonoles, Michael J" To: "Rouquette, Nicolas F (313K)" , "Nerijus Jankevicius" CC: "safriedenthal@gmail.com" , "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwCAASwrgIAABi2AgAAFSQCAAAPVgIAAGy3w Date: Tue, 28 May 2013 20:05:04 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [158.186.156.88] X-Proofpoint-Virus-Version: vendor=fsecure engine=2.50.10432:5.10.8626,1.0.431,0.0.0000 definitions=2013-05-28_08:2013-05-28,2013-05-28,1970-01-01 signatures=0 X-Virus-Scanned: amavisd-new at omg.org I.m a bit unclear, to take a UML DataType and make it ValeuType, just requires a stereotype? If so, that.s a lot of gratuitous work for the sake of formality. I have many type libraries developed (and used) by UML models and they are now starting to be used by SysML models. I don.t have the ability to edit the original, I don.t want to have two set of these, and would be concerned that a datatype enumeration of traffic light colors wouldn.t be compatible with the valuetype enumeration as they are different declarations. Michael From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Tuesday, May 28, 2013 10:21 AM To: Nerijus Jankevicius Cc: Chonoles, Michael J; safriedenthal@gmail.com; sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Nerijus, I understood your question; the answer is still yes. SysML ValueType extends DataType. This means ValueType can be applied to any kind of UML DataType, including specializations; I.e., PrimitiveType and Enumeration. Note that a value property in SysML must be typed by a SysML ValueType, even if it is a PrimitiveType or Enumeration. A property typed by a non-ValueType PrimitiveType or Enumeration is not, by definition, a value property in SysML. Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 7:07 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas, You misunderstood my question. I'm asking about usage of UML primitive types in SysML models. Is this restricted or not? If yes, where in SysML spec? I used VerdictKind as example. Nerijus On May 28, 2013, at 4:48 PM, Rouquette, Nicolas F (313K) wrote: Yes; ValueType can be applied to a PrimitiveType and to an Enumeration; both are kinds of DataType. You already filed an issue about VerdictKind missing the ValueType stereotype -- http://www.omg.org/issues/sysml-rtf.html#Issue18312 Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 6:26 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Does SysML allow typing operation parameters, behavior parameters and pins by UML primitive types? Can we use Enumerations which are not stereotyped by ValueType (like VerdictKind is) ? If no, where it is described and constrained? Thanks. Nerijus On May 28, 2013, at 2:31 AM, Rouquette, Nicolas F (313K) wrote: Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Rouquette, Nicolas F (313K)" To: "Chonoles, Michael J" , Nerijus Jankevicius CC: "safriedenthal@gmail.com" , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwCAAV52gP//kM+AgAB6pgD//455AIAA1Y4A//+sB4A= Date: Tue, 28 May 2013 22:04:35 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: user-agent: Microsoft-MacOutlook/14.3.4.130416 x-originating-ip: [128.149.137.113] X-Source-Sender: nicolas.f.rouquette@jpl.nasa.gov X-AUTH: Authorized X-Virus-Scanned: amavisd-new at omg.org Michael, Currently, ProfileApplication is a directed relationship that is owned by the applying Package (see Figure 12.12 in UML 2.5). To use in SysML a library originally written for UML only, it is currently necessary to modify this library to apply the SysML profile and record the stereotype instances such as those corresponding to the application of SysML ValueType to the UML DataTypes defined in that library. If we cannot modify the original library, then we cannot apply the SysML profile. In that case, one would have to duplicate the library and apply the SysML profile to the copy. This is clearly undesirable. To relax this problem, it would be necessary to relax the ownership of ProfileApplication and change the semantics of Profile such that instances of Stereotypes applied to elements would be associated to the applying Package. In this example, the UML library would remain as-is; there would be a new SysML library that would import the UML library and this SysML library would own a ProfileApplication for the SysML Profile. Then, the SysML library would have the metadata corresponding to all of the instances of SysML stereotypes applied to elements defined in the SysML library or accessible via importation. Would this address your problem? - Nicolas. From: , Michael Chonoles Date: Tuesday, May 28, 2013 1:05 PM To: JPL , Nerijus Jankevicius Cc: Sanford Friedenthal , "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete I.m a bit unclear, to take a UML DataType and make it ValeuType, just requires a stereotype? If so, that.s a lot of gratuitous work for the sake of formality. I have many type libraries developed (and used) by UML models and they are now starting to be used by SysML models. I don.t have the ability to edit the original, I don.t want to have two set of these, and would be concerned that a datatype enumeration of traffic light colors wouldn.t be compatible with the valuetype enumeration as they are different declarations. Michael From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Tuesday, May 28, 2013 10:21 AM To: Nerijus Jankevicius Cc: Chonoles, Michael J; safriedenthal@gmail.com; sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Nerijus, I understood your question; the answer is still yes. SysML ValueType extends DataType. This means ValueType can be applied to any kind of UML DataType, including specializations; I.e., PrimitiveType and Enumeration. Note that a value property in SysML must be typed by a SysML ValueType, even if it is a PrimitiveType or Enumeration. A property typed by a non-ValueType PrimitiveType or Enumeration is not, by definition, a value property in SysML. Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 7:07 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas, You misunderstood my question. I'm asking about usage of UML primitive types in SysML models. Is this restricted or not? If yes, where in SysML spec? I used VerdictKind as example. Nerijus On May 28, 2013, at 4:48 PM, Rouquette, Nicolas F (313K) wrote: Yes; ValueType can be applied to a PrimitiveType and to an Enumeration; both are kinds of DataType. You already filed an issue about VerdictKind missing the ValueType stereotype -- http://www.omg.org/issues/sysml-rtf.html#Issue18312 Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 6:26 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Does SysML allow typing operation parameters, behavior parameters and pins by UML primitive types? Can we use Enumerations which are not stereotyped by ValueType (like VerdictKind is) ? If no, where it is described and constrained? Thanks. Nerijus On May 28, 2013, at 2:31 AM, Rouquette, Nicolas F (313K) wrote: Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Chonoles, Michael J" To: "Rouquette, Nicolas F (313K)" , "Nerijus Jankevicius" CC: "safriedenthal@gmail.com" , "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwCAASwrgIAABi2AgAAFSQCAAAPVgIAAGy3wgABmaID//72LoA== Date: Tue, 28 May 2013 22:13:03 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [158.186.156.88] X-Proofpoint-Virus-Version: vendor=fsecure engine=2.50.10432:5.10.8626,1.0.431,0.0.0000 definitions=2013-05-28_09:2013-05-28,2013-05-28,1970-01-01 signatures=0 X-Virus-Scanned: amavisd-new at omg.org Well, it would go a long way, but I.m still afraid that a UML datatype and a System Valuetype, otherwise the same, wouldn.t work well together. We often have mixed models For example, we have pre-existing message definitions that are used for 1) System-To-System Interface Definitions 2) SysML Models 3) UML Models 4) Code Generation The message definitions are done in UML. Even for tracability reasons, we can.t have separate UML and SysML versions of these. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Tuesday, May 28, 2013 6:05 PM To: Chonoles, Michael J; Nerijus Jankevicius Cc: safriedenthal@gmail.com; sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Currently, ProfileApplication is a directed relationship that is owned by the applying Package (see Figure 12.12 in UML 2.5). To use in SysML a library originally written for UML only, it is currently necessary to modify this library to apply the SysML profile and record the stereotype instances such as those corresponding to the application of SysML ValueType to the UML DataTypes defined in that library. If we cannot modify the original library, then we cannot apply the SysML profile. In that case, one would have to duplicate the library and apply the SysML profile to the copy. This is clearly undesirable. To relax this problem, it would be necessary to relax the ownership of ProfileApplication and change the semantics of Profile such that instances of Stereotypes applied to elements would be associated to the applying Package. In this example, the UML library would remain as-is; there would be a new SysML library that would import the UML library and this SysML library would own a ProfileApplication for the SysML Profile. Then, the SysML library would have the metadata corresponding to all of the instances of SysML stereotypes applied to elements defined in the SysML library or accessible via importation. Would this address your problem? - Nicolas. From: , Michael Chonoles Date: Tuesday, May 28, 2013 1:05 PM To: JPL , Nerijus Jankevicius Cc: Sanford Friedenthal , "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete I.m a bit unclear, to take a UML DataType and make it ValeuType, just requires a stereotype? If so, that.s a lot of gratuitous work for the sake of formality. I have many type libraries developed (and used) by UML models and they are now starting to be used by SysML models. I don.t have the ability to edit the original, I don.t want to have two set of these, and would be concerned that a datatype enumeration of traffic light colors wouldn.t be compatible with the valuetype enumeration as they are different declarations. Michael From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Tuesday, May 28, 2013 10:21 AM To: Nerijus Jankevicius Cc: Chonoles, Michael J; safriedenthal@gmail.com; sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Nerijus, I understood your question; the answer is still yes. SysML ValueType extends DataType. This means ValueType can be applied to any kind of UML DataType, including specializations; I.e., PrimitiveType and Enumeration. Note that a value property in SysML must be typed by a SysML ValueType, even if it is a PrimitiveType or Enumeration. A property typed by a non-ValueType PrimitiveType or Enumeration is not, by definition, a value property in SysML. Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 7:07 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas, You misunderstood my question. I'm asking about usage of UML primitive types in SysML models. Is this restricted or not? If yes, where in SysML spec? I used VerdictKind as example. Nerijus On May 28, 2013, at 4:48 PM, Rouquette, Nicolas F (313K) wrote: Yes; ValueType can be applied to a PrimitiveType and to an Enumeration; both are kinds of DataType. You already filed an issue about VerdictKind missing the ValueType stereotype -- http://www.omg.org/issues/sysml-rtf.html#Issue18312 Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 6:26 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Does SysML allow typing operation parameters, behavior parameters and pins by UML primitive types? Can we use Enumerations which are not stereotyped by ValueType (like VerdictKind is) ? If no, where it is described and constrained? Thanks. Nerijus On May 28, 2013, at 2:31 AM, Rouquette, Nicolas F (313K) wrote: Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Rouquette, Nicolas F (313K)" To: "Chonoles, Michael J" , Nerijus Jankevicius CC: "safriedenthal@gmail.com" , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwCAAV52gP//kM+AgAB6pgD//455AIAA1Y4A//+sB4AADvdlgP//l7GA Date: Tue, 28 May 2013 22:59:45 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: user-agent: Microsoft-MacOutlook/14.3.4.130416 x-originating-ip: [128.149.137.113] X-Source-Sender: nicolas.f.rouquette@jpl.nasa.gov X-AUTH: Authorized X-Virus-Scanned: amavisd-new at omg.org Michael, In principle, UML/SysML supports what you want. Let's say that P1 is your UML library of DataTypes, PrimitiveTypes and Enumerations. Say, P1::DT, P1::PT and P1::ET are, respectively, a UML DataType, PrimitiveType and Enumeration. Let's say that P2 is a package that imports P1. P2 applies the SysML profile. In the context of P2, we should be able to apply SysML::ValueType to P2's imported elements . e.g., P1::DT, P1::PT and P1::ET When saving this, P1 shouldn't be modified. The information about the SysML stereotype instances applied to P2's imported elements should be stored in P2, not P1. Is this what you want? - Nicolas. From: , Michael Chonoles Date: Tuesday, May 28, 2013 3:13 PM To: JPL , Nerijus Jankevicius Cc: Sanford Friedenthal , "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Well, it would go a long way, but I.m still afraid that a UML datatype and a System Valuetype, otherwise the same, wouldn.t work well together. We often have mixed models For example, we have pre-existing message definitions that are used for 1) System-To-System Interface Definitions 2) SysML Models 3) UML Models 4) Code Generation The message definitions are done in UML. Even for tracability reasons, we can.t have separate UML and SysML versions of these. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Tuesday, May 28, 2013 6:05 PM To: Chonoles, Michael J; Nerijus Jankevicius Cc: safriedenthal@gmail.com; sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Currently, ProfileApplication is a directed relationship that is owned by the applying Package (see Figure 12.12 in UML 2.5). To use in SysML a library originally written for UML only, it is currently necessary to modify this library to apply the SysML profile and record the stereotype instances such as those corresponding to the application of SysML ValueType to the UML DataTypes defined in that library. If we cannot modify the original library, then we cannot apply the SysML profile. In that case, one would have to duplicate the library and apply the SysML profile to the copy. This is clearly undesirable. To relax this problem, it would be necessary to relax the ownership of ProfileApplication and change the semantics of Profile such that instances of Stereotypes applied to elements would be associated to the applying Package. In this example, the UML library would remain as-is; there would be a new SysML library that would import the UML library and this SysML library would own a ProfileApplication for the SysML Profile. Then, the SysML library would have the metadata corresponding to all of the instances of SysML stereotypes applied to elements defined in the SysML library or accessible via importation. Would this address your problem? - Nicolas. From: , Michael Chonoles Date: Tuesday, May 28, 2013 1:05 PM To: JPL , Nerijus Jankevicius Cc: Sanford Friedenthal , "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete I.m a bit unclear, to take a UML DataType and make it ValeuType, just requires a stereotype? If so, that.s a lot of gratuitous work for the sake of formality. I have many type libraries developed (and used) by UML models and they are now starting to be used by SysML models. I don.t have the ability to edit the original, I don.t want to have two set of these, and would be concerned that a datatype enumeration of traffic light colors wouldn.t be compatible with the valuetype enumeration as they are different declarations. Michael From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Tuesday, May 28, 2013 10:21 AM To: Nerijus Jankevicius Cc: Chonoles, Michael J; safriedenthal@gmail.com; sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Nerijus, I understood your question; the answer is still yes. SysML ValueType extends DataType. This means ValueType can be applied to any kind of UML DataType, including specializations; I.e., PrimitiveType and Enumeration. Note that a value property in SysML must be typed by a SysML ValueType, even if it is a PrimitiveType or Enumeration. A property typed by a non-ValueType PrimitiveType or Enumeration is not, by definition, a value property in SysML. Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 7:07 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas, You misunderstood my question. I'm asking about usage of UML primitive types in SysML models. Is this restricted or not? If yes, where in SysML spec? I used VerdictKind as example. Nerijus On May 28, 2013, at 4:48 PM, Rouquette, Nicolas F (313K) wrote: Yes; ValueType can be applied to a PrimitiveType and to an Enumeration; both are kinds of DataType. You already filed an issue about VerdictKind missing the ValueType stereotype -- http://www.omg.org/issues/sysml-rtf.html#Issue18312 Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 6:26 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Does SysML allow typing operation parameters, behavior parameters and pins by UML primitive types? Can we use Enumerations which are not stereotyped by ValueType (like VerdictKind is) ? If no, where it is described and constrained? Thanks. Nerijus On May 28, 2013, at 2:31 AM, Rouquette, Nicolas F (313K) wrote: Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Chonoles, Michael J" To: "Rouquette, Nicolas F (313K)" , "Nerijus Jankevicius" CC: "safriedenthal@gmail.com" , "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwCAASwrgIAABi2AgAAFSQCAAAPVgIAAGy3wgABmaID//72LoIAAUd+AgAB2azA= Date: Wed, 29 May 2013 10:11:03 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: x-originating-ip: [158.186.156.88] X-Proofpoint-Virus-Version: vendor=fsecure engine=2.50.10432:5.10.8626,1.0.431,0.0.0000 definitions=2013-05-29_04:2013-05-28,2013-05-29,1970-01-01 signatures=0 X-Virus-Scanned: amavisd-new at omg.org If were. able the SysML::ValueType en masse, that is without dealing individually with hundred.s of fields, I.d be a long way. But I also have messages or signals built on these Datatypes 1) Do I need to modify them or redo them to receive/send them w/I SysML 2) If I do have make changes, and send a modified message/signal, can a UML model handle them correctly (e.g., thinking of activity to activity diagram communication . either call behavior from a SysML activity diagram to lower-level UML operation, or a SysML send message to a UML wait for message) Michael From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Tuesday, May 28, 2013 7:00 PM To: Chonoles, Michael J; Nerijus Jankevicius Cc: safriedenthal@gmail.com; sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, In principle, UML/SysML supports what you want. Let's say that P1 is your UML library of DataTypes, PrimitiveTypes and Enumerations. Say, P1::DT, P1::PT and P1::ET are, respectively, a UML DataType, PrimitiveType and Enumeration. Let's say that P2 is a package that imports P1. P2 applies the SysML profile. In the context of P2, we should be able to apply SysML::ValueType to P2's imported elements . e.g., P1::DT, P1::PT and P1::ET When saving this, P1 shouldn't be modified. The information about the SysML stereotype instances applied to P2's imported elements should be stored in P2, not P1. Is this what you want? - Nicolas. From: , Michael Chonoles Date: Tuesday, May 28, 2013 3:13 PM To: JPL , Nerijus Jankevicius Cc: Sanford Friedenthal , "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Well, it would go a long way, but I.m still afraid that a UML datatype and a System Valuetype, otherwise the same, wouldn.t work well together. We often have mixed models For example, we have pre-existing message definitions that are used for 1) System-To-System Interface Definitions 2) SysML Models 3) UML Models 4) Code Generation The message definitions are done in UML. Even for tracability reasons, we can.t have separate UML and SysML versions of these. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Tuesday, May 28, 2013 6:05 PM To: Chonoles, Michael J; Nerijus Jankevicius Cc: safriedenthal@gmail.com; sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Currently, ProfileApplication is a directed relationship that is owned by the applying Package (see Figure 12.12 in UML 2.5). To use in SysML a library originally written for UML only, it is currently necessary to modify this library to apply the SysML profile and record the stereotype instances such as those corresponding to the application of SysML ValueType to the UML DataTypes defined in that library. If we cannot modify the original library, then we cannot apply the SysML profile. In that case, one would have to duplicate the library and apply the SysML profile to the copy. This is clearly undesirable. To relax this problem, it would be necessary to relax the ownership of ProfileApplication and change the semantics of Profile such that instances of Stereotypes applied to elements would be associated to the applying Package. In this example, the UML library would remain as-is; there would be a new SysML library that would import the UML library and this SysML library would own a ProfileApplication for the SysML Profile. Then, the SysML library would have the metadata corresponding to all of the instances of SysML stereotypes applied to elements defined in the SysML library or accessible via importation. Would this address your problem? - Nicolas. From: , Michael Chonoles Date: Tuesday, May 28, 2013 1:05 PM To: JPL , Nerijus Jankevicius Cc: Sanford Friedenthal , "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete I.m a bit unclear, to take a UML DataType and make it ValeuType, just requires a stereotype? If so, that.s a lot of gratuitous work for the sake of formality. I have many type libraries developed (and used) by UML models and they are now starting to be used by SysML models. I don.t have the ability to edit the original, I don.t want to have two set of these, and would be concerned that a datatype enumeration of traffic light colors wouldn.t be compatible with the valuetype enumeration as they are different declarations. Michael From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Tuesday, May 28, 2013 10:21 AM To: Nerijus Jankevicius Cc: Chonoles, Michael J; safriedenthal@gmail.com; sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Nerijus, I understood your question; the answer is still yes. SysML ValueType extends DataType. This means ValueType can be applied to any kind of UML DataType, including specializations; I.e., PrimitiveType and Enumeration. Note that a value property in SysML must be typed by a SysML ValueType, even if it is a PrimitiveType or Enumeration. A property typed by a non-ValueType PrimitiveType or Enumeration is not, by definition, a value property in SysML. Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 7:07 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas, You misunderstood my question. I'm asking about usage of UML primitive types in SysML models. Is this restricted or not? If yes, where in SysML spec? I used VerdictKind as example. Nerijus On May 28, 2013, at 4:48 PM, Rouquette, Nicolas F (313K) wrote: Yes; ValueType can be applied to a PrimitiveType and to an Enumeration; both are kinds of DataType. You already filed an issue about VerdictKind missing the ValueType stereotype -- http://www.omg.org/issues/sysml-rtf.html#Issue18312 Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 6:26 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Does SysML allow typing operation parameters, behavior parameters and pins by UML primitive types? Can we use Enumerations which are not stereotyped by ValueType (like VerdictKind is) ? If no, where it is described and constrained? Thanks. Nerijus On May 28, 2013, at 2:31 AM, Rouquette, Nicolas F (313K) wrote: Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas. From: "Rouquette, Nicolas F (313K)" To: "Chonoles, Michael J" , Nerijus Jankevicius CC: "safriedenthal@gmail.com" , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Topic: EXTERNAL: RE: The SysML classification of properties is incomplete Thread-Index: AQHOUAVFT5MwThY260SYdHAAA483EJkZFF1ggABGIYCAAEFGAIAAAJYwgAAoSwCAAV52gP//kM+AgAB6pgD//455AIAA1Y4A//+sB4AADvdlgP//l7GAgAEw6oD//84wAA== Date: Wed, 29 May 2013 14:12:49 +0000 Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: user-agent: Microsoft-MacOutlook/14.3.4.130416 x-originating-ip: [128.149.137.113] X-Source-Sender: nicolas.f.rouquette@jpl.nasa.gov X-AUTH: Authorized X-Virus-Scanned: amavisd-new at omg.org Again, in principle, I believe it would work. In practice, make some test cases and check with your tool vendor. - Nicolas. From: , Michael Chonoles Date: Wednesday, May 29, 2013 3:11 AM To: JPL , Nerijus Jankevicius Cc: Sanford Friedenthal , "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If were. able the SysML::ValueType en masse, that is without dealing individually with hundred.s of fields, I.d be a long way. But I also have messages or signals built on these Datatypes 1) Do I need to modify them or redo them to receive/send them w/I SysML 2) If I do have make changes, and send a modified message/signal, can a UML model handle them correctly (e.g., thinking of activity to activity diagram communication . either call behavior from a SysML activity diagram to lower-level UML operation, or a SysML send message to a UML wait for message) Michael From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Tuesday, May 28, 2013 7:00 PM To: Chonoles, Michael J; Nerijus Jankevicius Cc: safriedenthal@gmail.com; sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, In principle, UML/SysML supports what you want. Let's say that P1 is your UML library of DataTypes, PrimitiveTypes and Enumerations. Say, P1::DT, P1::PT and P1::ET are, respectively, a UML DataType, PrimitiveType and Enumeration. Let's say that P2 is a package that imports P1. P2 applies the SysML profile. In the context of P2, we should be able to apply SysML::ValueType to P2's imported elements . e.g., P1::DT, P1::PT and P1::ET When saving this, P1 shouldn't be modified. The information about the SysML stereotype instances applied to P2's imported elements should be stored in P2, not P1. Is this what you want? - Nicolas. From: , Michael Chonoles Date: Tuesday, May 28, 2013 3:13 PM To: JPL , Nerijus Jankevicius Cc: Sanford Friedenthal , "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Well, it would go a long way, but I.m still afraid that a UML datatype and a System Valuetype, otherwise the same, wouldn.t work well together. We often have mixed models For example, we have pre-existing message definitions that are used for 1) System-To-System Interface Definitions 2) SysML Models 3) UML Models 4) Code Generation The message definitions are done in UML. Even for tracability reasons, we can.t have separate UML and SysML versions of these. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Tuesday, May 28, 2013 6:05 PM To: Chonoles, Michael J; Nerijus Jankevicius Cc: safriedenthal@gmail.com; sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Currently, ProfileApplication is a directed relationship that is owned by the applying Package (see Figure 12.12 in UML 2.5). To use in SysML a library originally written for UML only, it is currently necessary to modify this library to apply the SysML profile and record the stereotype instances such as those corresponding to the application of SysML ValueType to the UML DataTypes defined in that library. If we cannot modify the original library, then we cannot apply the SysML profile. In that case, one would have to duplicate the library and apply the SysML profile to the copy. This is clearly undesirable. To relax this problem, it would be necessary to relax the ownership of ProfileApplication and change the semantics of Profile such that instances of Stereotypes applied to elements would be associated to the applying Package. In this example, the UML library would remain as-is; there would be a new SysML library that would import the UML library and this SysML library would own a ProfileApplication for the SysML Profile. Then, the SysML library would have the metadata corresponding to all of the instances of SysML stereotypes applied to elements defined in the SysML library or accessible via importation. Would this address your problem? - Nicolas. From: , Michael Chonoles Date: Tuesday, May 28, 2013 1:05 PM To: JPL , Nerijus Jankevicius Cc: Sanford Friedenthal , "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete I.m a bit unclear, to take a UML DataType and make it ValeuType, just requires a stereotype? If so, that.s a lot of gratuitous work for the sake of formality. I have many type libraries developed (and used) by UML models and they are now starting to be used by SysML models. I don.t have the ability to edit the original, I don.t want to have two set of these, and would be concerned that a datatype enumeration of traffic light colors wouldn.t be compatible with the valuetype enumeration as they are different declarations. Michael From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Tuesday, May 28, 2013 10:21 AM To: Nerijus Jankevicius Cc: Chonoles, Michael J; safriedenthal@gmail.com; sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Nerijus, I understood your question; the answer is still yes. SysML ValueType extends DataType. This means ValueType can be applied to any kind of UML DataType, including specializations; I.e., PrimitiveType and Enumeration. Note that a value property in SysML must be typed by a SysML ValueType, even if it is a PrimitiveType or Enumeration. A property typed by a non-ValueType PrimitiveType or Enumeration is not, by definition, a value property in SysML. Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 7:07 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas, You misunderstood my question. I'm asking about usage of UML primitive types in SysML models. Is this restricted or not? If yes, where in SysML spec? I used VerdictKind as example. Nerijus On May 28, 2013, at 4:48 PM, Rouquette, Nicolas F (313K) wrote: Yes; ValueType can be applied to a PrimitiveType and to an Enumeration; both are kinds of DataType. You already filed an issue about VerdictKind missing the ValueType stereotype -- http://www.omg.org/issues/sysml-rtf.html#Issue18312 Nicolas. From: Nerijus Jankevicius Date: Tuesday, May 28, 2013 6:26 AM To: JPL Cc: Michael Chonoles , Sanford Friedenthal , "sysml-rtf@omg.org" Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Does SysML allow typing operation parameters, behavior parameters and pins by UML primitive types? Can we use Enumerations which are not stereotyped by ValueType (like VerdictKind is) ? If no, where it is described and constrained? Thanks. Nerijus On May 28, 2013, at 2:31 AM, Rouquette, Nicolas F (313K) wrote: Michael, Yes, you can type an operation or behavior parameter with a block. The issue I raised is that, regardless of modeling methodology, the SysML spec is currently unclear as to whether it is allowed to type a signal property by a block or not. Assuming that this is just a matter of clarification in the spec, then modeling methodology becomes a relevant matter. I hope you recognize the importance of maintaining the separation between blocks that model something about the structure of the system (which can be dynamically evolving) vs. something about dynamic objects involved in the operation of that system. - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 2:11 PM To: JPL , Sanford Friedenthal Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete If I can send a block via an operation parameters, I should be able to send a block by a signal. The use of an operation vs signal is mostly a difference in mechanism and synch. Even if methodology, use of a block via signal is sometime dangerous, which I.m not sure I agree with, I don.t what SysML to enforce your methodology. From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 27, 2013 5:05 PM To: Chonoles, Michael J; safriedenthal@gmail.com Cc: sysml-rtf@omg.org Subject: Re: EXTERNAL: RE: The SysML classification of properties is incomplete Michael, Your statement is grammatically incorrect. Can you please clarify your position? - Nicolas. From: , Michael Chonoles Date: Monday, May 27, 2013 10:13 AM To: Sanford Friedenthal , JPL Cc: "sysml-rtf@omg.org" Subject: RE: EXTERNAL: RE: The SysML classification of properties is incomplete Yes, It.s important to have signal properties typed by blocks, as they could be have a unique identity . not just a value. Michael From: Sanford Friedenthal [mailto:safriedenthal@gmail.com] Sent: Monday, May 27, 2013 9:05 AM To: 'Rouquette, Nicolas F (313K)' Cc: sysml-rtf@omg.org Subject: EXTERNAL: RE: The SysML classification of properties is incomplete Nicolas I sometimes use blocks to type pins on actions to represent data flow between the actions. For example, one action may produce a data output called status, and send this to another action. This status may have more detailed data content, which may be typed by value properties if it is primitive type data, or it may be typed by blocks if it has further structure. If this status data is also used to trigger a transition on a state machine, it generally would correspond to the property of a signal. In this case, I would think we would want to allow signals to have properties typed by block. Your thoughts? Regards, Sandy From: Rouquette, Nicolas F (313K) [mailto:nicolas.f.rouquette@jpl.nasa.gov] Sent: Monday, May 13, 2013 2:11 PM To: issues@omg.org Cc: sysml-rtf@omg.org Subject: The SysML classification of properties is incomplete SysML 1.3 section 8.3.2.2 Block says: SysML establishes four basic classifications of properties belonging to a SysML Block or ValueType. . A property typed by a SysML ValueType is classified as a value property, and always has composite aggregation. In SysML, we also have Signals. In UML, Signals can have properties. How does SysML then classify properties defined in a Signal? A very strict reading of the SysML spec would suggest that a Signal cannot have any kind of SysML property because a Signal is neither a SysML Block nor a SysML ValueType. However, this is clearly too restrictive in practice. I propose expanding SysML's classification of properties to include SysML Blocks, ValueTypes and Signals. This leads to another question: What are the legal types of a property belonging to a Signal? I propose restricting such properties to be typed by SysML ValueTypes only. This corresponds to the practical situation where a Signal carries a data payload . I.e., it is a message with some data content. Allowing a property belonging to a Signal to be typed by a SysML Block or Signal leads to semantic problems . what would it mean to send / receive such signals? - Nicolas.