Issue 14552: Ordered derived unions (uml2-rtf) Source: International Business Machines (Dr. Maged Elaasar, melaasar(at)ca.ibm.com) Nature: Uncategorized Issue Severity: Summary: When you specify a property as an ordered derived union, with the implied derivation of union-ing the subsetting properties, how do you specify the order of the resulting union? Resolution: Revised Text: Actions taken: October 7, 2009: received issue October 9, 2009: moved from MOF Core RTF to the UML 2 Infrastructure RTF Discussion: End of Annotations:===== ubject: Ordered derived unions To: uml2-rtf@omg.org X-Mailer: Lotus Notes Release 7.0 HF277 June 21, 2006 From: Maged Elaasar Date: Wed, 7 Oct 2009 17:03:57 -0400 X-MIMETrack: Serialize by Router on D25ML03/25/M/IBM(Release 8.0.1|February 07, 2008) at 10/07/2009 17:03:58 When you specify a property as an ordered derived union, with the implied derivation of union-ing the subsetting properties, how do you specify the order of the resulting union? Maged Elaasar, PhD Candidate Senior Software Engineer, Rational Modeling Tools IBM Representative@OMG, CAS Research Staff Member IBM Canada, Ottawa Lab, +1 613 270 4651 Subject: RE: Ordered derived unions Date: Wed, 7 Oct 2009 14:13:48 -0700 X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: Ordered derived unions Thread-Index: AcpHkbjxrfD4/PxVQlKR/4OvGvDy/AAAGFag From: "Pete Rivett" To: "Maged Elaasar" , There is no such capability provided. There is in fact no need to specify the order of the resulting union . all that is meant by {ordered} is that the order is maintained and not randomly different on subsequent accesses to the property value (assuming no intervening update). I.m not sure how much sense {ordered} makes for a {union}. It minimally requires all the subsetting properties to also be {ordered} . we should either specify that constraint or disallow it. Pete From: Maged Elaasar [mailto:melaasar@ca.ibm.com] Sent: 07 October 2009 14:04 To: uml2-rtf@omg.org Subject: Ordered derived unions When you specify a property as an ordered derived union, with the implied derivation of union-ing the subsetting properties, how do you specify the order of the resulting union? Maged Elaasar, PhD Candidate Senior Software Engineer, Rational Modeling Tools IBM Representative@OMG, CAS Research Staff Member IBM Canada, Ottawa Lab, +1 613 270 4651 Subject: RE: Ordered derived unions Date: Thu, 8 Oct 2009 11:51:23 -0400 X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: Ordered derived unions thread-index: AQHKR5IliipxxxMFyEeAq6iPRSJFD5D6hjcAgADZieCAAG8OkA== From: "Ed Seidewitz" To: "Steve Cook" , "Pete Rivett - Adaptive" , "Maged Elaasar" , Note that, in the UML action metamodel, the add structural feature value action has an .insertAt. pin for ordered features (which essentially has the behavior of .insert before., with a value of .*. meaning .append at end.). Of course, this isn.t available for MOF. -- Ed -------------------------------------------------------------------------------- From: Steve Cook [mailto:Steve.Cook@microsoft.com] Sent: Thursday, October 08, 2009 5:20 AM To: Pete Rivett - Adaptive; Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions >all that is meant by {ordered} is that the order is maintained and not randomly different on subsequent accesses to the property value (assuming no intervening update). I was surprised to read that, but on looking at the MOF spec it appears to be true. I also expect an ordered property to be one in which it is possible to control where insertions occur, i.e. the API includes operations like insertAfter, insertBefore, etc. Otherwise how is it possible to create a decent modeling tool! From: Pete Rivett [mailto:pete.rivett@adaptive.com] Sent: 07 October 2009 22:14 To: Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions There is no such capability provided. There is in fact no need to specify the order of the resulting union . all that is meant by {ordered} is that the order is maintained and not randomly different on subsequent accesses to the property value (assuming no intervening update). I.m not sure how much sense {ordered} makes for a {union}. It minimally requires all the subsetting properties to also be {ordered} . we should either specify that constraint or disallow it. Pete From: Maged Elaasar [mailto:melaasar@ca.ibm.com] Sent: 07 October 2009 14:04 To: uml2-rtf@omg.org Subject: Ordered derived unions When you specify a property as an ordered derived union, with the implied derivation of union-ing the subsetting properties, how do you specify the order of the resulting union? Maged Elaasar, PhD Candidate Senior Software Engineer, Rational Modeling Tools IBM Representative@OMG, CAS Research Staff Member IBM Canada, Ottawa Lab, +1 613 270 4651 Subject: RE: Ordered derived unions Date: Thu, 8 Oct 2009 09:11:48 -0700 X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: Ordered derived unions Thread-Index: AQHKR5IliipxxxMFyEeAq6iPRSJFD5D6hjcAgADZieCAAG8OkIAAA/Xg From: "Pete Rivett" To: "Ed Seidewitz" , "Steve Cook" , "Maged Elaasar" , MOF does define interfaces for maintaining order: specifically 10.7 ReflectiveSequence has add(index, object), remove(index), set(index, object). That was not my point which was not about controlling the order of the content but about .specifying. the order.. Pete From: Ed Seidewitz [mailto:ed-s@modeldriven.com] Sent: 08 October 2009 08:51 To: Steve Cook; Pete Rivett; Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions Note that, in the UML action metamodel, the add structural feature value action has an .insertAt. pin for ordered features (which essentially has the behavior of .insert before., with a value of .*. meaning .append at end.). Of course, this isn.t available for MOF. -- Ed -------------------------------------------------------------------------------- From: Steve Cook [mailto:Steve.Cook@microsoft.com] Sent: Thursday, October 08, 2009 5:20 AM To: Pete Rivett - Adaptive; Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions >all that is meant by {ordered} is that the order is maintained and not randomly different on subsequent accesses to the property value (assuming no intervening update). I was surprised to read that, but on looking at the MOF spec it appears to be true. I also expect an ordered property to be one in which it is possible to control where insertions occur, i.e. the API includes operations like insertAfter, insertBefore, etc. Otherwise how is it possible to create a decent modeling tool! From: Pete Rivett [mailto:pete.rivett@adaptive.com] Sent: 07 October 2009 22:14 To: Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions There is no such capability provided. There is in fact no need to specify the order of the resulting union . all that is meant by {ordered} is that the order is maintained and not randomly different on subsequent accesses to the property value (assuming no intervening update). I.m not sure how much sense {ordered} makes for a {union}. It minimally requires all the subsetting properties to also be {ordered} . we should either specify that constraint or disallow it. Pete From: Maged Elaasar [mailto:melaasar@ca.ibm.com] Sent: 07 October 2009 14:04 To: uml2-rtf@omg.org Subject: Ordered derived unions When you specify a property as an ordered derived union, with the implied derivation of union-ing the subsetting properties, how do you specify the order of the resulting union? Maged Elaasar, PhD Candidate Senior Software Engineer, Rational Modeling Tools IBM Representative@OMG, CAS Research Staff Member IBM Canada, Ottawa Lab, +1 613 270 4651 From: Steve Cook To: Pete Rivett , Ed Seidewitz , Maged Elaasar , "uml2-rtf@omg.org" Subject: RE: Ordered derived unions Thread-Topic: Ordered derived unions Thread-Index: AQHKR5IliipxxxMFyEeAq6iPRSJFD5D6hjcAgADZieCAAG8OkIAAA/XggAAGctA= Date: Thu, 8 Oct 2009 16:30:54 +0000 Accept-Language: en-GB, en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: Ah yes: I stand corrected. I was actually looking at the CMOF semantics chapter, which talks about links, but says nothing about how they might be ordered with respect to their ends. Perhaps that semantics chapter should be deleted. I doubt it adds significant value. -- Steve From: Pete Rivett [mailto:pete.rivett@adaptive.com] Sent: 08 October 2009 17:12 To: Ed Seidewitz; Steve Cook; Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions MOF does define interfaces for maintaining order: specifically 10.7 ReflectiveSequence has add(index, object), remove(index), set(index, object). That was not my point which was not about controlling the order of the content but about .specifying. the order.. Pete From: Ed Seidewitz [mailto:ed-s@modeldriven.com] Sent: 08 October 2009 08:51 To: Steve Cook; Pete Rivett; Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions Note that, in the UML action metamodel, the add structural feature value action has an .insertAt. pin for ordered features (which essentially has the behavior of .insert before., with a value of .*. meaning .append at end.). Of course, this isn.t available for MOF. -- Ed -------------------------------------------------------------------------------- From: Steve Cook [mailto:Steve.Cook@microsoft.com] Sent: Thursday, October 08, 2009 5:20 AM To: Pete Rivett - Adaptive; Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions >all that is meant by {ordered} is that the order is maintained and not randomly different on subsequent accesses to the property value (assuming no intervening update). I was surprised to read that, but on looking at the MOF spec it appears to be true. I also expect an ordered property to be one in which it is possible to control where insertions occur, i.e. the API includes operations like insertAfter, insertBefore, etc. Otherwise how is it possible to create a decent modeling tool! From: Pete Rivett [mailto:pete.rivett@adaptive.com] Sent: 07 October 2009 22:14 To: Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions There is no such capability provided. There is in fact no need to specify the order of the resulting union . all that is meant by {ordered} is that the order is maintained and not randomly different on subsequent accesses to the property value (assuming no intervening update). I.m not sure how much sense {ordered} makes for a {union}. It minimally requires all the subsetting properties to also be {ordered} . we should either specify that constraint or disallow it. Pete From: Maged Elaasar [mailto:melaasar@ca.ibm.com] Sent: 07 October 2009 14:04 To: uml2-rtf@omg.org Subject: Ordered derived unions When you specify a property as an ordered derived union, with the implied derivation of union-ing the subsetting properties, how do you specify the order of the resulting union? Maged Elaasar, PhD Candidate Senior Software Engineer, Rational Modeling Tools IBM Representative@OMG, CAS Research Staff Member IBM Canada, Ottawa Lab, +1 613 270 4651 Subject: RE: Ordered derived unions To: "Pete Rivett" Cc: "Ed Seidewitz" , "Steve Cook" , uml2-rtf@omg.org X-Mailer: Lotus Notes Release 7.0 HF277 June 21, 2006 From: Maged Elaasar Date: Thu, 8 Oct 2009 12:31:23 -0400 X-MIMETrack: Serialize by Router on D25ML03/25/M/IBM(Release 8.0.1|February 07, 2008) at 10/08/2009 12:31:23 Yes - since ordered derived unions are "derived", you can only control the order through "specification", which is not a supported feature as you mentioned. "Pete Rivett" "Pete Rivett" 10/08/2009 12:11 PM To "Ed Seidewitz" , "Steve Cook" , Maged Elaasar/Ottawa/IBM@IBMCA, cc Subject RE: Ordered derived unions MOF does define interfaces for maintaining order: specifically 10.7 ReflectiveSequence has add(index, object), remove(index), set(index, object). That was not my point which was not about controlling the order of the content but about .specifying. the order.. Pete From: Ed Seidewitz [mailto:ed-s@modeldriven.com] Sent: 08 October 2009 08:51 To: Steve Cook; Pete Rivett; Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions Note that, in the UML action metamodel, the add structural feature value action has an .insertAt. pin for ordered features (which essentially has the behavior of .insert before., with a value of .*. meaning .append at end.). Of course, this isn.t available for MOF. -- Ed -------------------------------------------------------------------------------- From: Steve Cook [mailto:Steve.Cook@microsoft.com] Sent: Thursday, October 08, 2009 5:20 AM To: Pete Rivett - Adaptive; Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions >all that is meant by {ordered} is that the order is maintained and not randomly different on subsequent accesses to the property value (assuming no intervening update). I was surprised to read that, but on looking at the MOF spec it appears to be true. I also expect an ordered property to be one in which it is possible to control where insertions occur, i.e. the API includes operations like insertAfter, insertBefore, etc. Otherwise how is it possible to create a decent modeling tool! From: Pete Rivett [mailto:pete.rivett@adaptive.com] Sent: 07 October 2009 22:14 To: Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions There is no such capability provided. There is in fact no need to specify the order of the resulting union . all that is meant by {ordered} is that the order is maintained and not randomly different on subsequent accesses to the property value (assuming no intervening update). I.m not sure how much sense {ordered} makes for a {union}. It minimally requires all the subsetting properties to also be {ordered} . we should either specify that constraint or disallow it. Pete From: Maged Elaasar [mailto:melaasar@ca.ibm.com] Sent: 07 October 2009 14:04 To: uml2-rtf@omg.org Subject: Ordered derived unions When you specify a property as an ordered derived union, with the implied derivation of union-ing the subsetting properties, how do you specify the order of the resulting union? Maged Elaasar, PhD Candidate Senior Software Engineer, Rational Modeling Tools IBM Representative@OMG, CAS Research Staff Member IBM Canada, Ottawa Lab, +1 613 270 4651 pic32629.gif From: "Bock, Conrad" To: "uml2-rtf@omg.org" Date: Thu, 8 Oct 2009 12:43:14 -0400 Subject: RE: Ordered derived unions Thread-Topic: Ordered derived unions Thread-Index: AcpINQXgP5XLRSVxSjSqDRwql/ptcQAAL0+A Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: acceptlanguage: en-US X-NIST-MailScanner: Found to be clean X-NIST-MailScanner-From: conrad.bock@nist.gov X-MIME-Autoconverted: from quoted-printable to 8bit by amethyst.omg.org id n98GfiqP019522 Maged, > Yes - since ordered derived unions are "derived", you can > only control the order through "specification", which is > not a supported feature as you mentioned. I hope derivation is different than read-only. You should be able to set a derived property as long as the value is consistent with the derivation (or the derivation is reversable). In this case, you should be able to reorder the values as long as order specified in the subsets (if any) is preserved in the union. There was some discussion of interchanging derived values that are set, but I don't remember the outcome. Conrad From: Steve Cook To: "Bock, Conrad" , "uml2-rtf@omg.org" Subject: RE: Ordered derived unions Thread-Topic: Ordered derived unions Thread-Index: AQHKR5IliipxxxMFyEeAq6iPRSJFD5D6hjcAgADZieCAAG8OkIAAA/Xg///24YCAAANPAIAAE7nQ Date: Thu, 8 Oct 2009 16:54:00 +0000 Accept-Language: en-GB, en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: X-MIME-Autoconverted: from quoted-printable to 8bit by amethyst.omg.org id n98GqX88021699 >From UML: [8] A derived union is read only. isDerivedUnion implies isReadOnly -----Original Message----- From: Bock, Conrad [mailto:conrad.bock@nist.gov] Sent: 08 October 2009 17:43 To: uml2-rtf@omg.org Subject: RE: Ordered derived unions Maged, > Yes - since ordered derived unions are "derived", you can > only control the order through "specification", which is > not a supported feature as you mentioned. I hope derivation is different than read-only. You should be able to set a derived property as long as the value is consistent with the derivation (or the derivation is reversable). In this case, you should be able to reorder the values as long as order specified in the subsets (if any) is preserved in the union. There was some discussion of interchanging derived values that are set, but I don't remember the outcome. Conrad From: "Bock, Conrad" To: "uml2-rtf@omg.org" Date: Thu, 8 Oct 2009 12:57:23 -0400 Subject: RE: Ordered derived unions Thread-Topic: Ordered derived unions Thread-Index: AQHKR5IliipxxxMFyEeAq6iPRSJFD5D6hjcAgADZieCAAG8OkIAAA/Xg///24YCAAANPAIAAE7nQgAAA6kA= Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: acceptlanguage: en-US X-NIST-MailScanner: Found to be clean X-NIST-MailScanner-From: conrad.bock@nist.gov X-MIME-Autoconverted: from quoted-printable to 8bit by amethyst.omg.org id n98GtsHi022373 Steve, > [8] A derived union is read only. > isDerivedUnion implies isReadOnly Bug. :) Conrad Subject: RE: Ordered derived unions Date: Fri, 9 Oct 2009 08:54:06 +0200 X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: Ordered derived unions Thread-Index: AQHKR5IliipxxxMFyEeAq6iPRSJFD5D6hjcAgADZieCAAG8OkIAAA/Xg///24YCAAANPAIAAE7nQgAAA6kCAAOMZoA== From: "BERNARD, Yves" To: "Bock, Conrad" , X-OriginalArrivalTime: 09 Oct 2009 06:54:07.0266 (UTC) FILETIME=[4BDAE820:01CA48AD] X-MIME-Autoconverted: from quoted-printable to 8bit by amethyst.omg.org id n996qZbb027089 Conrad, I disagree. By definition a "derived" property has its value calculated according to value from other properties. If its not the case, it means that the property is not derived. On the other hand, a non-derived property may have a default value calculated (i.e. derived) from value of other properties, but its only its default value... To me, if there is an inconsistency somewhere it because all derived properties should be readonly. Consider the practical impact of this statement from the UML specification (§7.3.44, semantic section): "where a derived property is changeable, an implementation is expected to make appropriate changes to the model in order for all the constraints to be met, in particular the derivation constraint for the derived property." More generaly, I don't think that derived union can be actually ordered because an union is a commutative law. Yves -----Message d'origine----- De : Bock, Conrad [mailto:conrad.bock@nist.gov] Envoyé : jeudi 8 octobre 2009 18:57 À : uml2-rtf@omg.org Objet : RE: Ordered derived unions Steve, > [8] A derived union is read only. > isDerivedUnion implies isReadOnly Bug. :) Conrad This mail has originated outside your organization, either from an external partner or the Global Internet. Keep this in mind if you answer this message. The information in this e-mail is confidential. The contents may not be disclosed or used by anyone other than the addressee. Access to this e-mail by anyone else is unauthorised. If you are not the intended recipient, please notify Airbus immediately and delete this e-mail. Airbus cannot accept any responsibility for the accuracy or completeness of this e-mail as it has been sent over public networks. If you have any concerns over the content of this message or its Accuracy or Integrity, please contact Airbus immediately. All outgoing e-mails from Airbus are checked using regularly updated virus scanning software but you should take whatever measures you deem to be appropriate to ensure that this message and any attachments are virus free. Date: Fri, 09 Oct 2009 11:48:54 +0100 From: Dave Hawkins User-Agent: Thunderbird 2.0.0.6 (X11/20070728) To: "Bock, Conrad" CC: "uml2-rtf@omg.org" Subject: Re: Ordered derived unions X-Source-IP: acsmt353.oracle.com [141.146.40.153] X-Auth-Type: Internal IP X-CT-RefId: str=0001.0A090207.4ACF151E.0134:SCFMA4539814,ss=1,fgs=0 Conrad, Bock, Conrad wrote: Maged, > Yes - since ordered derived unions are "derived", you can > only control the order through "specification", which is > not a supported feature as you mentioned. I hope derivation is different than read-only. You should be able to set a derived property as long as the value is consistent with the derivation (or the derivation is reversable). In this case, you should be able to reorder the values as long as order specified in the subsets (if any) is preserved in the union. There was some discussion of interchanging derived values that are set, but I don't remember the outcome. If the union has to preserve the order of its subsets that would imply the subsets are all ordered in the same way. That could be regarded as the meaning of an ordered union. However I'm not sure how useful that would actually be and I think it would be better that unions never be ordered. If you want to explicitly order the union, then, as Bernard has said, the property is no longer derived because you are providing information that cannot be derived from other properties. It's not strictly the same as a non-derived property though because it's constrained to be the union of its subsets. So I could see an argument there for not requiring a union to be derived. Writeable derived properties are a nasty feature that I don't think should be allowed. They should always have the correct values anyway because they are derived from other properties, so in general it should never be necessary to set them. Setting them also requires reverse derivations, which are not supplied in the UML metamodel. I have a feeling writeable derived properties were allowed to support the various compliance levels, because a property could become derived in a higher compliance level. Thus to XMI import a lower compliance level in a higher level would require 'setting' the derived property. If compliance levels are now being removed, it should be considered whether derived properties should be anything other than read-only. Dave -- Dave Hawkins | Principal Software Engineer | +44 118 924 0022 Oracle JDeveloper Development Oracle Corporation UK Ltd is a company incorporated in England & Wales. Company Reg. No. 1782505. Reg. office: Oracle Parkway, Thames Valley Park, Reading RG6 1RA. Subject: RE: Ordered derived unions Date: Fri, 9 Oct 2009 08:15:32 -0700 X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: Ordered derived unions Thread-Index: AcpIzkwtGU3juwtERBy+6EltaZFrfgAJC0gg From: "Pete Rivett" To: "Dave Hawkins" Cc: , "Bock, Conrad" X-MIME-Autoconverted: from quoted-printable to 8bit by amethyst.omg.org id n99FE7Nq019009 > I have a feeling writeable derived properties were allowed to support the various compliance levels, because a property could become derived in a higher > compliance level. Thus to XMI import a lower compliance level in a higher level would require 'setting' the derived property. > If compliance levels are now being removed, it should be considered whether derived properties should be anything other than read-only. Writeable derived properties are useful in 'normal' modeling where there are 2 properties which are representations of the same value - for example 'weightInPounds' derived from 'weightInKilos'. Setting the former would implicitly set the other. Another example is something like 'fullName' derived from a number of other properties like 'firstNames', 'lastName' etc. Setting fullName could set the other attributes given a sufficiently sophisticated parser. Likewise 'fullAddress'. Pete -----Original Message----- From: Dave Hawkins [mailto:dave.hawkins@oracle.com] Sent: 09 October 2009 03:49 To: Bock, Conrad Cc: uml2-rtf@omg.org Subject: Re: Ordered derived unions Conrad, Bock, Conrad wrote: > Maged, > > > Yes - since ordered derived unions are "derived", you can > > only control the order through "specification", which is > > not a supported feature as you mentioned. > > I hope derivation is different than read-only. You should be able to > set a derived property as long as the value is consistent with the > derivation (or the derivation is reversable). In this case, you should > be able to reorder the values as long as order specified in the subsets > (if any) is preserved in the union. There was some discussion of > interchanging derived values that are set, but I don't remember the > outcome. If the union has to preserve the order of its subsets that would imply the subsets are all ordered in the same way. That could be regarded as the meaning of an ordered union. However I'm not sure how useful that would actually be and I think it would be better that unions never be ordered. If you want to explicitly order the union, then, as Bernard has said, the property is no longer derived because you are providing information that cannot be derived from other properties. It's not strictly the same as a non-derived property though because it's constrained to be the union of its subsets. So I could see an argument there for not requiring a union to be derived. Writeable derived properties are a nasty feature that I don't think should be allowed. They should always have the correct values anyway because they are derived from other properties, so in general it should never be necessary to set them. Setting them also requires reverse derivations, which are not supplied in the UML metamodel. I have a feeling writeable derived properties were allowed to support the various compliance levels, because a property could become derived in a higher compliance level. Thus to XMI import a lower compliance level in a higher level would require 'setting' the derived property. If compliance levels are now being removed, it should be considered whether derived properties should be anything other than read-only. Dave -- Dave Hawkins | Principal Software Engineer | +44 118 924 0022 Oracle JDeveloper Development Oracle Corporation UK Ltd is a company incorporated in England & Wales. Company Reg. No. 1782505. Reg. office: Oracle Parkway, Thames Valley Park, Reading RG6 1RA. From: "Bock, Conrad" To: "uml2-rtf@omg.org" Date: Fri, 9 Oct 2009 11:22:41 -0400 Subject: RE: Ordered derived unions Thread-Topic: Ordered derived unions Thread-Index: AcpIzi6pomJD2VFmSRi8WlPMURhU8wAJaaAw Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: acceptlanguage: en-US X-NIST-MailScanner: Found to be clean X-NIST-MailScanner-From: conrad.bock@nist.gov X-MIME-Autoconverted: from quoted-printable to 8bit by amethyst.omg.org id n99FL52u020506 Dave, > If the union has to preserve the order of its subsets that would > imply the subsets are all ordered in the same way. No, it just means the order of elements in the union that come from the same susbet are in the same order as the subset. The subsets don't need to be ordered by the same criteria. > Writeable derived properties are a nasty feature that I don't think > should be allowed. They should always have the correct values anyway > because they are derived from other properties, so in general it > should never be necessary to set them. Setting them also requires > reverse derivations, which are not supplied in the UML metamodel. The values the derivation is based on might not be there. For example, you should be able to interchange the area of a square even if the width and height aren't known. Conrad From: "Bock, Conrad" To: "BERNARD, Yves" , "uml2-rtf@omg.org" Date: Fri, 9 Oct 2009 11:25:05 -0400 Subject: RE: Ordered derived unions Thread-Topic: Ordered derived unions Thread-Index: AQHKR5IliipxxxMFyEeAq6iPRSJFD5D6hjcAgADZieCAAG8OkIAAA/Xg///24YCAAANPAIAAE7nQgAAA6kCAAOMZoIAAlOMA Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: acceptlanguage: en-US X-NIST-MailScanner: Found to be clean X-NIST-MailScanner-From: conrad.bock@nist.gov X-MIME-Autoconverted: from quoted-printable to 8bit by amethyst.omg.org id n99FNQw1020973 Yves, > I disagree. By definition a "derived" property has its > value calculated according to value from other properties. > If its not the case, it means that the property is not > derived. On the other hand, a non-derived property may have > a default value calculated (i.e. derived) from value of > other properties, but its only its default value... > To me, if there is an inconsistency somewhere it because > all derived properties should be readonly. Consider the > practical impact of this statement from the UML > specification (§7.3.44, semantic section): > "where a derived property is changeable, an implementation > is expected to make appropriate changes to the model in > order for all the constraints to be met, in particular the > derivation constraint for the derived property." See my message to Dave. It's very useful to treat derived properties as specified by constraint, so explicitly writing a derived value places a constraint on the values used in the derivation. Conrad X-Mailer: QUALCOMM Windows Eudora Version 7.1.0.9 Date: Fri, 09 Oct 2009 11:41:35 -0400 To: Steve Cook , Pete Rivett , Ed Seidewitz , Maged Elaasar , "uml2-rtf@omg.org" From: Juergen Boldt Subject: RE: Ordered derived unions should I assign an issue number to this email thread? -Juergen At 12:30 PM 10/8/2009, Steve Cook wrote: Ah yes: I stand corrected. I was actually looking at the CMOF semantics chapter, which talks about links, but says nothing about how they might be ordered with respect to their ends. Perhaps that semantics chapter should be deleted. I doubt it adds significant value. -- Steve From: Pete Rivett [ mailto:pete.rivett@adaptive.com] Sent: 08 October 2009 17:12 To: Ed Seidewitz; Steve Cook; Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions MOF does define interfaces for maintaining order: specifically 10.7 ReflectiveSequence has add(index, object), remove(index), set(index, object). That was not my point which was not about controlling the order of the content but about .specifying. the order.. Pete From: Ed Seidewitz [ mailto:ed-s@modeldriven.com] Sent: 08 October 2009 08:51 To: Steve Cook; Pete Rivett; Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions Note that, in the UML action metamodel, the add structural feature value action has an .insertAt. pin for ordered features (which essentially has the behavior of .insert before., with a value of .*. meaning .append at end.). Of course, this isn.t available for MOF. -- Ed -------------------------------------------------------------------------------- From: Steve Cook [ mailto:Steve.Cook@microsoft.com] Sent: Thursday, October 08, 2009 5:20 AM To: Pete Rivett - Adaptive; Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions >all that is meant by {ordered} is that the order is maintained and not randomly different on subsequent accesses to the property value (assuming no intervening update). I was surprised to read that, but on looking at the MOF spec it appears to be true. I also expect an ordered property to be one in which it is possible to control where insertions occur, i.e. the API includes operations like insertAfter, insertBefore, etc. Otherwise how is it possible to create a decent modeling tool! From: Pete Rivett [ mailto:pete.rivett@adaptive.com] Sent: 07 October 2009 22:14 To: Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions There is no such capability provided. There is in fact no need to specify the order of the resulting union . all that is meant by {ordered} is that the order is maintained and not randomly different on subsequent accesses to the property value (assuming no intervening update). I.m not sure how much sense {ordered} makes for a {union}. It minimally requires all the subsetting properties to also be {ordered} . we should either specify that constraint or disallow it. Pete From: Maged Elaasar [ mailto:melaasar@ca.ibm.com] Sent: 07 October 2009 14:04 To: uml2-rtf@omg.org Subject: Ordered derived unions When you specify a property as an ordered derived union, with the implied derivation of union-ing the subsetting properties, how do you specify the order of the resulting union? Maged Elaasar, PhD Candidate Senior Software Engineer, Rational Modeling Tools IBM Representative@OMG, CAS Research Staff Member IBM Canada, Ottawa Lab, +1 613 270 4651 Juergen Boldt Director, Member Services Object Management Group 140 Kendrick St Building A Suite 300 Needham, MA 02494 USA tel: +1 781 444 0404 x 132 fax: +1 781 444 0320 email: juergen@omg.org www.omg.org From: Steve Cook To: Juergen Boldt , Pete Rivett , Ed Seidewitz , Maged Elaasar , "uml2-rtf@omg.org" Subject: RE: Ordered derived unions Thread-Topic: Ordered derived unions Thread-Index: AQHKR5IliipxxxMFyEeAq6iPRSJFD5D6hjcAgADZieCAAG8OkIAAA/XggAAGctCAAYZZVoAAATxg Date: Fri, 9 Oct 2009 15:49:46 +0000 Accept-Language: en-GB, en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: Juergen I think it is a MOF issue: CMOF semantics chapter says nothing about link ordering. So yes please. Thanks -- Steve From: Juergen Boldt [mailto:juergen@omg.org] Sent: 09 October 2009 16:42 To: Steve Cook; Pete Rivett; Ed Seidewitz; Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions should I assign an issue number to this email thread? -Juergen At 12:30 PM 10/8/2009, Steve Cook wrote: Ah yes: I stand corrected. I was actually looking at the CMOF semantics chapter, which talks about links, but says nothing about how they might be ordered with respect to their ends. Perhaps that semantics chapter should be deleted. I doubt it adds significant value. -- Steve From: Pete Rivett [ mailto:pete.rivett@adaptive.com] Sent: 08 October 2009 17:12 To: Ed Seidewitz; Steve Cook; Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions MOF does define interfaces for maintaining order: specifically 10.7 ReflectiveSequence has add(index, object), remove(index), set(index, object). That was not my point which was not about controlling the order of the content but about .specifying. the order.. Pete From: Ed Seidewitz [ mailto:ed-s@modeldriven.com] Sent: 08 October 2009 08:51 To: Steve Cook; Pete Rivett; Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions Note that, in the UML action metamodel, the add structural feature value action has an .insertAt. pin for ordered features (which essentially has the behavior of .insert before., with a value of .*. meaning .append at end.). Of course, this isn.t available for MOF. -- Ed -------------------------------------------------------------------------------- From: Steve Cook [ mailto:Steve.Cook@microsoft.com] Sent: Thursday, October 08, 2009 5:20 AM To: Pete Rivett - Adaptive; Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions >all that is meant by {ordered} is that the order is maintained and not randomly different on subsequent accesses to the property value (assuming no intervening update). I was surprised to read that, but on looking at the MOF spec it appears to be true. I also expect an ordered property to be one in which it is possible to control where insertions occur, i.e. the API includes operations like insertAfter, insertBefore, etc. Otherwise how is it possible to create a decent modeling tool! From: Pete Rivett [ mailto:pete.rivett@adaptive.com] Sent: 07 October 2009 22:14 To: Maged Elaasar; uml2-rtf@omg.org Subject: RE: Ordered derived unions There is no such capability provided. There is in fact no need to specify the order of the resulting union . all that is meant by {ordered} is that the order is maintained and not randomly different on subsequent accesses to the property value (assuming no intervening update). I.m not sure how much sense {ordered} makes for a {union}. It minimally requires all the subsetting properties to also be {ordered} . we should either specify that constraint or disallow it. Pete From: Maged Elaasar [ mailto:melaasar@ca.ibm.com] Sent: 07 October 2009 14:04 To: uml2-rtf@omg.org Subject: Ordered derived unions When you specify a property as an ordered derived union, with the implied derivation of union-ing the subsetting properties, how do you specify the order of the resulting union? Maged Elaasar, PhD Candidate Senior Software Engineer, Rational Modeling Tools IBM Representative@OMG, CAS Research Staff Member IBM Canada, Ottawa Lab, +1 613 270 4651 Juergen Boldt Director, Member Services Object Management Group 140 Kendrick St Building A Suite 300 Needham, MA 02494 USA tel: +1 781 444 0404 x 132 fax: +1 781 444 0320 email: juergen@omg.org www.omg.org X-Mailer: QUALCOMM Windows Eudora Version 7.1.0.9 Date: Fri, 09 Oct 2009 12:00:15 -0400 To: issues@omg.org, mof2core-rtf@omg.org From: Juergen Boldt Subject: issue 14552 -- MOF 2 Core RTF issue Cc: uml2-rtf@omg.org Subject: Ordered derived unions To: uml2-rtf@omg.org X-Mailer: Lotus Notes Release 7.0 HF277 June 21, 2006 From: Maged Elaasar Date: Wed, 7 Oct 2009 17:03:57 -0400 X-MIMETrack: Serialize by Router on D25ML03/25/M/IBM(Release 8.0.1|February 07, 2008) at 10/07/2009 17:03:58 When you specify a property as an ordered derived union, with the implied derivation of union-ing the subsetting properties, how do you specify the order of the resulting union? Maged Elaasar, PhD Candidate Senior Software Engineer, Rational Modeling Tools IBM Representative@OMG, CAS Research Staff Member IBM Canada, Ottawa Lab, +1 613 270 4651 Juergen Boldt Director, Member Services Object Management Group 140 Kendrick St Building A Suite 300 Needham, MA 02494 USA tel: +1 781 444 0404 x 132 fax: +1 781 444 0320 email: juergen@omg.org www.omg.org From: "Bock, Conrad" To: "uml2-rtf@omg.org" Date: Fri, 9 Oct 2009 12:05:57 -0400 Subject: RE: Ordered derived unions Thread-Topic: Ordered derived unions Thread-Index: AQHKR5IliipxxxMFyEeAq6iPRSJFD5D6hjcAgADZieCAAG8OkIAAA/XggAAGctCAAYZZVoAAATxggAABzpA= Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: acceptlanguage: en-US X-NIST-MailScanner: Found to be clean X-NIST-MailScanner-From: conrad.bock@nist.gov X-MIME-Autoconverted: from quoted-printable to 8bit by amethyst.omg.org id n99G4HRc028734 Steve, > I think it is a MOF issue: CMOF semantics chapter says > nothing about link ordering. So yes please. Is this shared with UML? In any case, link ordering is different than this thread one setting derived properties, how to specify ordering in derived unions, etc, though related of course. Conrad From: Steve Cook To: "Bock, Conrad" , "uml2-rtf@omg.org" Subject: RE: Ordered derived unions Thread-Topic: Ordered derived unions Thread-Index: AQHKR5IliipxxxMFyEeAq6iPRSJFD5D6hjcAgADZieCAAG8OkIAAA/XggAAGctCAAYZZVoAAATxggAABzpCAAAOToA== Date: Fri, 9 Oct 2009 16:08:46 +0000 Accept-Language: en-GB, en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: X-MIME-Autoconverted: from quoted-printable to 8bit by amethyst.omg.org id n99G7L8K029155 Not shared with UML. MOF has its own semantics chapter. -----Original Message----- From: Bock, Conrad [mailto:conrad.bock@nist.gov] Sent: 09 October 2009 17:06 To: uml2-rtf@omg.org Subject: RE: Ordered derived unions Steve, > I think it is a MOF issue: CMOF semantics chapter says > nothing about link ordering. So yes please. Is this shared with UML? In any case, link ordering is different than this thread one setting derived properties, how to specify ordering in derived unions, etc, though related of course. Conrad From: "Bock, Conrad" To: "uml2-rtf@omg.org" Date: Fri, 9 Oct 2009 12:12:11 -0400 Subject: RE: Ordered derived unions Thread-Topic: Ordered derived unions Thread-Index: AQHKR5IliipxxxMFyEeAq6iPRSJFD5D6hjcAgADZieCAAG8OkIAAA/XggAAGctCAAYZZVoAAATxggAABzpCAAAOToIAAANrg Accept-Language: en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: acceptlanguage: en-US X-NIST-MailScanner: Found to be clean X-NIST-MailScanner-From: conrad.bock@nist.gov X-MIME-Autoconverted: from quoted-printable to 8bit by amethyst.omg.org id n99GAXUX029583 Steve, > Not shared with UML. MOF has its own semantics chapter. In that case, I'd reassign the issue Juergen just entered to UML/Infrastructure, because UML has derived uonions also, then file another issue about links in the MOF semantics. Conrad Subject: RE: issue 14552 -- MOF 2 Core RTF issue Date: Fri, 9 Oct 2009 09:39:10 -0700 X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: issue 14552 -- MOF 2 Core RTF issue Thread-Index: AQHKSPoTKnVpSKi23UOnDIKK5oJCO5D9aUfQgAABbGA= From: "Pete Rivett" To: "Steve Cook" Cc: Ø In fact the MOF semantics chapter seems to add no value and perhaps it should be deleted from the spec altogether. Ø Notwithstanding the omission or ordering, are you saying there is inherently .no value. in attempting to represent the specific semantics of MOF, or that the current attempt to do that fails? If the latter, is it because of the approach taken (specifying the primitive operations in terms of post conditions at the instance level) or the current implementation of that approach? If the approach is the problem, what approach would you advocate? Pete From: Steve Cook [mailto:Steve.Cook@microsoft.com] Sent: 09 October 2009 09:31 To: Juergen Boldt; issues@omg.org; mof2core-rtf@omg.org Cc: uml2-rtf@omg.org Subject: RE: issue 14552 -- MOF 2 Core RTF issue Juergen Can you reassign this one to UML infrastructure please, and raise a different issue against MOF: MOF semantics chapter says nothing about ordering of links when association ends are marked .ordered.. In fact the MOF semantics chapter seems to add no value and perhaps it should be deleted from the spec altogether. Thanks - -Steve From: Juergen Boldt [mailto:juergen@omg.org] Sent: 09 October 2009 17:00 To: issues@omg.org; mof2core-rtf@omg.org Cc: uml2-rtf@omg.org Subject: issue 14552 -- MOF 2 Core RTF issue Subject: Ordered derived unions To: uml2-rtf@omg.org X-Mailer: Lotus Notes Release 7.0 HF277 June 21, 2006 From: Maged Elaasar Date: Wed, 7 Oct 2009 17:03:57 -0400 X-MIMETrack: Serialize by Router on D25ML03/25/M/IBM(Release 8.0.1|February 07, 2008) at 10/07/2009 17:03:58 When you specify a property as an ordered derived union, with the implied derivation of union-ing the subsetting properties, how do you specify the order of the resulting union? Maged Elaasar, PhD Candidate Senior Software Engineer, Rational Modeling Tools IBM Representative@OMG, CAS Research Staff Member IBM Canada, Ottawa Lab, +1 613 270 4651 Juergen Boldt Director, Member Services Object Management Group 140 Kendrick St Building A Suite 300 Needham, MA 02494 USA tel: +1 781 444 0404 x 132 fax: +1 781 444 0320 email: juergen@omg.org www.omg.org From: Steve Cook To: Pete Rivett CC: "mof2core-rtf@omg.org" Subject: RE: issue 14552 -- MOF 2 Core RTF issue Thread-Topic: issue 14552 -- MOF 2 Core RTF issue Thread-Index: AQHKSPoTKnVpSKi23UOnDIKK5oJCO5D9aUfQgAABbGCAAAQUcA== Date: Fri, 9 Oct 2009 16:52:01 +0000 Accept-Language: en-GB, en-US X-MS-Has-Attach: X-MS-TNEF-Correlator: There just seems to be plenty of relatively easily digestible information in the rest of the spec that says what the operations do, whereas the semantics chapter is verbose, and verification that what it says is meaningful is intractable. From: Pete Rivett [mailto:pete.rivett@adaptive.com] Sent: 09 October 2009 17:39 To: Steve Cook Cc: mof2core-rtf@omg.org Subject: RE: issue 14552 -- MOF 2 Core RTF issue Ø In fact the MOF semantics chapter seems to add no value and perhaps it should be deleted from the spec altogether. Ø Notwithstanding the omission or ordering, are you saying there is inherently .no value. in attempting to represent the specific semantics of MOF, or that the current attempt to do that fails? If the latter, is it because of the approach taken (specifying the primitive operations in terms of post conditions at the instance level) or the current implementation of that approach? If the approach is the problem, what approach would you advocate? Pete From: Steve Cook [mailto:Steve.Cook@microsoft.com] Sent: 09 October 2009 09:31 To: Juergen Boldt; issues@omg.org; mof2core-rtf@omg.org Cc: uml2-rtf@omg.org Subject: RE: issue 14552 -- MOF 2 Core RTF issue Juergen Can you reassign this one to UML infrastructure please, and raise a different issue against MOF: MOF semantics chapter says nothing about ordering of links when association ends are marked .ordered.. In fact the MOF semantics chapter seems to add no value and perhaps it should be deleted from the spec altogether. Thanks - -Steve From: Juergen Boldt [mailto:juergen@omg.org] Sent: 09 October 2009 17:00 To: issues@omg.org; mof2core-rtf@omg.org Cc: uml2-rtf@omg.org Subject: issue 14552 -- MOF 2 Core RTF issue Subject: Ordered derived unions To: uml2-rtf@omg.org X-Mailer: Lotus Notes Release 7.0 HF277 June 21, 2006 From: Maged Elaasar Date: Wed, 7 Oct 2009 17:03:57 -0400 X-MIMETrack: Serialize by Router on D25ML03/25/M/IBM(Release 8.0.1|February 07, 2008) at 10/07/2009 17:03:58 When you specify a property as an ordered derived union, with the implied derivation of union-ing the subsetting properties, how do you specify the order of the resulting union? Maged Elaasar, PhD Candidate Senior Software Engineer, Rational Modeling Tools IBM Representative@OMG, CAS Research Staff Member IBM Canada, Ottawa Lab, +1 613 270 4651 Juergen Boldt Director, Member Services Object Management Group 140 Kendrick St Building A Suite 300 Needham, MA 02494 USA tel: +1 781 444 0404 x 132 fax: +1 781 444 0320 email: juergen@omg.org www.omg.org