Issue 5785: CORBA-WSDL/SOAP: Section 1.2.2.1 (corbawsdl-ftf) Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: Section 1.2.2.1 - tns is a single namespace for all mapped IDL from the entire corba world. The module names for IDL across this vast namespace, need to be sufficiently qualified with pre-fixes (such as Type ID, or Pragma Prefix) to avoid module name collisions. Proposed Resolution: FTF needs to determine appropriate prefixes or namespaces to assure module name uniqueness in the generated WSDL Resolution: see above Revised Text: Remove the following text from the “tns” row of the table in 1.2.2.1 “ (There are still some issues to be addressed at the FTF stage around how we should map repository IDs into namespaces, so it is likely that some sort of concatenation of namespace and ID will occur). “ Also, in section 1.2.4 Modules: Change “ IDL modules are represented in WSDL as prefixes to the names contained in the module. These prefixes are concatenated using a “.” as a separator. For example, an entity “X” in a module “M” will have the name “M.X” in WSDL. “ to the following: “ For every module name which does not collide with another module name, across the set of IDL artifacts which are being mapped to a WSDL document, then the scoped name of the module starting from the root module name using the "." character as separator between names shall be the effective Id used in the corresponding WSDL. If there is a module name collision, apply the following rules to the module name which causes the collision: a. If there are no pragma or typeid/typeprefix directives that applies to a module name then the scoped name of the module starting from the root module name using the "." character as separator between names shall be the effective Id used in the corresponding WSDL. b. If there is a pragma prefix or a typeprefix directive that applies to the module name then the prefix specified in the directive/pragma shall be of the form: <prefix string> "_" <effective Id from (1)> If no such directives apply to the module name in question then the effective Id shall be the same as in (1). c. If there is a pragma version that applies to the scopename then the version string shall be appened to the effective Id obtained in (2), that is the effective Id will be of the form: <effective Id from (2)> "_" <version string> d. If a pragma id or a typeid directive applies to the module name in question then: · If the first four characters of the string specified in the directive/pragma is "IDL:" then the string with the "IDL:" prefix removed and with the "/" characters in the remaining string substituted by the "." character shall be the effective Id. · If the Repository Id type prefix in the directive/pragma is something other than "IDL:" then the entire id string shall be used as the effective Id · If any characters which are not valid for XML element names are encountered in the mapped module name, they shall be replaced with “_”. Actions taken: December 17, 2002: received issue November 6, 2003: closed issue Discussion: Resolution: Append a prefix (if present and required) in front of the module name. Need to fix all examples to use this pragma prefix in module name prefixes. To avoid invalid XML element names, use “_” rather than “:” (which is reserved for use with XML namespaces) in the name prefixes End of Annotations:===== Date: Tue, 17 Dec 2002 12:30:41 -0500 From: Tom Rutt Reply-To: tom@coastin.com X-Mailer: Mozilla 4.78 [en] (Windows NT 5.0; U) X-Accept-Language: en To: juergen@omg.org, issues@omg.org, "corbawsdl-ftf@omg.org" Subject: MARS and AB FTF issues for CORBA-WSDL/SOAP interworking Juergen, could you please file these as official issues for the FTF to resolve. Thank you: Tom Rutt corbawsdl FTF chair --------------------- Initial FTF Issues on 02-10-12-draft-adopted-spec.PDF Issues from MARS and AB reviews: FTF Issue 1 Source: MARS FTF and AB review Section 1.2.2.1 - tns is a single namespace for all mapped IDL from the entire corba world. The module names for IDL across this vast namespace, need to be sufficiently qualified with pre-fixes (such as Type ID, or Pragma Prefix) to avoid module name collisions. Proposed Resolution: FTF needs to determine appropriate prefixes or namespaces to assure module name uniqueness in the generated WSDL. Date: Tue, 18 Mar 2003 20:18:42 -0500 From: Tom Rutt Reply-To: tom@coastin.com User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.0.2) Gecko/20021120 Netscape/7.01 X-Accept-Language: en-us, en To: "corbawsdl-ftf@omg.org" Subject: ISSUE 5785 Proposal X-OriginalArrivalTime: 19 Mar 2003 01:21:43.0171 (UTC) FILETIME=[E67E8530:01C2EDB5] Issue 5785: CORBA-WSDL/SOAP: Section 1.2.2.1 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: Section 1.2.2.1 - tns is a single namespace for all mapped IDL from the entire corba world. The module names for IDL across this vast namespace, need to be sufficiently qualified with pre-fixes (such as Type ID, or Pragma Prefix) to avoid module name collisions. Proposed Resolution: FTF needs to determine appropriate prefixes or namespaces to assure module name uniqueness in the generated WSDL Discussion: The brute force approach is to use the scheme that is used for IDL form of repository ID for a Module. Prefix the module name with the string from the pragma prefix IDL directive, usin g (i.e. An OMG IDL pragma of the format: #pragma prefi x ) Proposed Resolution: The WSDL name for the IDL Module should be patterned after the IDL form for the repository ID for the Module. Append the pragma prefix (if present, in front of the module name. If the module definition has no corresponding pragma prefix directive, the prefix string shall be empty. If pragma prefixes are not used, there could be collisions. If such collisions were detected in the mapped WSDL document, they could be resolved by the addition of an appropriate pragma prefix to one of the IDL module definitions. Need to fix all examples to use this pragma prefix in module name prefixes Proposed Revised Text: Remove the following text from th e row of the table in 1.2.2.1 (There are still some issues to be addressed at the FTF stage around how we should map repository IDs into namespaces, so it is likely that some sort of concatenation of namespace and ID will occur). Also, in section 1.2.4 Modules: Change IDL modules are represented in WSDL as prefixes to the names contained in the module. These prefixes are concatenated using a as a separator. For example, an entit y in a modul e will have the nam e in WSDL. to the following: IDL modules are represented in WSDL as module name prefixes to the names contained in the module. The module prefix to use shall be the pragma prefix associated with to the IDL module definition (or the empty string if there is no pragma prefix directive) concatenated.to the module name, using a as a separator. These module prefixes are concatenated using a as a separator. For example, an entit y in a modul e with pragma prefi x will have the nam e in WSDL. If there were no pragma prefix directive, the same IDL entit y in modul e would have the nam e in WSDL. Fix all examples to append the pragma prefix before the module name (at least pu t before the module name, if there is no pragma prefix in the IDL example). Actions taken: December 17, 2002: received issue -- ---------------------------------------------------- Tom Rutt email: tom@coastin.com; trutt@fsw.fujitsu.com Tel: +1 732 801 5744 Fax: +1 732 774 5133 Date: Wed, 19 Mar 2003 11:45:07 -0500 From: Tom Rutt Reply-To: tom@coastin.com User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.0.2) Gecko/20021120 Netscape/7.01 X-Accept-Language: en-us, en To: "corbawsdl-ftf@omg.org" CC: tom@coastin.com, iwasa Subject: Re: ISSUE 5785 Proposal X-OriginalArrivalTime: 19 Mar 2003 16:48:09.0218 (UTC) FILETIME=[525C2E20:01C2EE37] Upon further reflection, I think the proposal should be modified slightly. If there is no pragma prefix associated with the module, the module prefix should be the module name alone (without the ":" prefix). I think there might be problems with a name beginning with the ":" symbol, and also it is unnecessary. Thus the last example in the proposal should be "M.X" instead of ":M.X" Tom Rutt Tom Rutt wrote: Issue 5785: CORBA-WSDL/SOAP: Section 1.2.2.1 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: Section 1.2.2.1 - tns is a single namespace for all mapped IDL from the entire corba world. The module names for IDL across this vast namespace, need to be sufficiently qualified with pre-fixes (such as Type ID, or Pragma Prefix) to avoid module name collisions. Proposed Resolution: FTF needs to determine appropriate prefixes or namespaces to assure module name uniqueness in the generated WSDL Discussion: The brute force approach is to use the scheme that is used for IDL form of repository ID for a Module. Prefix the module name with the string from the pragma prefix IDL directive, usin g (i.e. An OMG IDL pragma of the format: #pragma prefi x ) Proposed Resolution: The WSDL name for the IDL Module should be patterned after the IDL form for the repository ID for the Module. Append the pragma prefix (if present, in front of the module name. If the module definition has no corresponding pragma prefix directive, the prefix string shall be empty. If pragma prefixes are not used, there could be collisions. If such collisions were detected in the mapped WSDL document, they could be resolved by the addition of an appropriate pragma prefix to one of the IDL module definitions. Need to fix all examples to use this pragma prefix in module name prefixes Proposed Revised Text: Remove the following text from th e row of the table in 1.2.2.1 Also, in section 1.2.4 Modules: Change. as a separator. For example, an entit y in a modul e will have the nam e in WSDL. to the following:: as a separator. These module prefixes are concatenated using a as a separator. For example, an entit y in a modul e with pragma prefi x will have the nam e in WSDL. If there were no pragma prefix directive, the same IDL entit y in modul e would have the nam e in WSDL. Fix all examples to append the pragma prefix before the module name (at least pu t before the module name, if there is no pragma prefix in the IDL example). Actions taken: December 17, 2002: received issue -- ---------------------------------------------------- Tom Rutt email: tom@coastin.com; trutt@fsw.fujitsu.com Tel: +1 732 801 5744 Fax: +1 732 774 5133 Date: Wed, 19 Mar 2003 12:59:20 -0500 From: Tom Rutt Reply-To: tom@coastin.com User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.0.2) Gecko/20021120 Netscape/7.01 X-Accept-Language: en-us, en To: Jishnu Mukerji CC: "corbawsdl-ftf@omg.org" , iwasa Subject: Re: ISSUE 5785 Proposal X-OriginalArrivalTime: 19 Mar 2003 18:02:21.0578 (UTC) FILETIME=[B02C66A0:01C2EE41] Another point: It will look like hell if every idl entity has a name starting with the IDL repository ID prefix. We are really doing a lot of extra prefixing to avoid using the typeID or orgID as a namespace rather than a name prefix. Tom Rutt Jishnu Mukerji wrote: In post 3.0 IDL there is no "pragma prefix". In post 3.0 IDL there is only "typeid". Also, pre 3.0 IDL one could use "pragma id" instead of "pragma prefix". Would it be the intention to use the typeid specified by "typeid" or "pragma id" if they are present too? If that is the case, why don't we simply say that the effective IDL Repository ID will be used, with all ":" by "."? In that case the general namespace name will be of the form "IDL.". Jishnu. Tom Rutt wrote: Upon further reflection, I think the proposal should be modified slightly. If there is no pragma prefix associated with the module, the module prefix should be the module name alone (without the ":" prefix). I think there might be problems with a name beginning with the ":" symbol, and also it is unnecessary. Thus the last example in the proposal should be "M.X" instead of ":M.X" Tom Rutt Tom Rutt wrote: Issue 5785: CORBA-WSDL/SOAP: Section 1.2.2.1 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: Section 1.2.2.1 - tns is a single namespace for all mapped IDL from the entire corba world. The module names for IDL across this vast namespace, need to be sufficiently qualified with pre-fixes (such as Type ID, or Pragma Prefix) to avoid module name collisions. Proposed Resolution: FTF needs to determine appropriate prefixes or namespaces to assure module name uniqueness in the generated WSDL Discussion: The brute force approach is to use the scheme that is used for IDL form of repository ID for a Module. Prefix the module name with the string from the pragma prefix IDL directive, usin g (i.e. An OMG IDL pragma of the format: #pragma prefi x ) Proposed Resolution: The WSDL name for the IDL Module should be patterned after the IDL form for the repository ID for the Module. Append the pragma prefix (if present, in front of the module name. If the module definition has no corresponding pragma prefix directive, the prefix string shall be empty. If pragma prefixes are not used, there could be collisions. If such collisions were detected in the mapped WSDL document, they could be resolved by the addition of an appropriate pragma prefix to one of the IDL module definitions. Need to fix all examples to use this pragma prefix in module name prefixes Proposed Revised Text: Remove the following text from th e row of the table in 1.2.2.1 (There are still some issues to be addressed at the FTF stage around how we should map repository IDs into namespaces, so it is likely that some sort of concatenation of namespace and ID will occur). Also, in section 1.2.4 Modules: Change IDL modules are represented in WSDL as prefixes to the names contained in the module. These prefixes are concatenated using a as a separator. For example, an entit y in a modul e will have the nam e in WSDL. to the following: IDL modules are represented in WSDL as module name prefixes to the names contained in the module. The module prefix to use shall be the pragma prefix associated with to the IDL module definition (or the empty string if there is no pragma prefix directive) concatenated.to the module name, using a as a separator. These module prefixes are concatenated using a as a separator. For example, an entit y in a modul e with pragma prefi x will have the nam e in WSDL. If there were no pragma prefix directive, the same IDL entit y in modul e would have the nam e in WSDL. Fix all examples to append the pragma prefix before the module name (at least pu t before the module name, if there is no pragma prefix in the IDL example). Actions taken: December 17, 2002: received issue -- ---------------------------------------------------- Tom Rutt email: tom@coastin.com; trutt@fsw.fujitsu.com Tel: +1 732 801 5744 Fax: +1 732 774 5133 Date: Wed, 19 Mar 2003 12:56:29 -0500 From: Tom Rutt Reply-To: tom@coastin.com User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.0.2) Gecko/20021120 Netscape/7.01 X-Accept-Language: en-us, en To: Jishnu Mukerji CC: "corbawsdl-ftf@omg.org" , iwasa Subject: Re: ISSUE 5785 Proposal X-OriginalArrivalTime: 19 Mar 2003 17:59:30.0281 (UTC) FILETIME=[4A128D90:01C2EE41] I thought about that, but UUID repository ids will be very ugly in the WSDL, for example. All we need it uniqueness, and adding the typeID or pragma prefix is enough to do that. Why add all the extra "machine looking" stuff like the IDL:, vs DCE: which have ugly numbers. Perhaps we should use some examples to see what they would look like. Tom Rutt Jishnu Mukerji wrote: In post 3.0 IDL there is no "pragma prefix". In post 3.0 IDL there is only "typeid". Also, pre 3.0 IDL one could use "pragma id" instead of "pragma prefix". Would it be the intention to use the typeid specified by "typeid" or "pragma id" if they are present too? If that is the case, why don't we simply say that the effective IDL Repository ID will be used, with all ":" by "."? In that case the general namespace name will be of the form "IDL.". Jishnu. Tom Rutt wrote: Upon further reflection, I think the proposal should be modified slightly. If there is no pragma prefix associated with the module, the module prefix should be the module name alone (without the ":" prefix). I think there might be problems with a name beginning with the ":" symbol, and also it is unnecessary. Thus the last example in the proposal should be "M.X" instead of ":M.X" Tom Rutt Tom Rutt wrote: Issue 5785: CORBA-WSDL/SOAP: Section 1.2.2.1 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: Section 1.2.2.1 - tns is a single namespace for all mapped IDL from the entire corba world. The module names for IDL across this vast namespace, need to be sufficiently qualified with pre-fixes (such as Type ID, or Pragma Prefix) to avoid module name collisions. Proposed Resolution: FTF needs to determine appropriate prefixes or namespaces to assure module name uniqueness in the generated WSDL Discussion: The brute force approach is to use the scheme that is used for IDL form of repository ID for a Module. Prefix the module name with the string from the pragma prefix IDL directive, usin g (i.e. An OMG IDL pragma of the format: #pragma prefi x ) Proposed Resolution: The WSDL name for the IDL Module should be patterned after the IDL form for the repository ID for the Module. Append the pragma prefix (if present, in front of the module name. If the module definition has no corresponding pragma prefix directive, the prefix string shall be empty. If pragma prefixes are not used, there could be collisions. If such collisions were detected in the mapped WSDL document, they could be resolved by the addition of an appropriate pragma prefix to one of the IDL module definitions. Need to fix all examples to use this pragma prefix in module name prefixes Proposed Revised Text: Remove the following text from th e row of the table in 1.2.2.1 (There are still some issues to be addressed at the FTF stage around how we should map repository IDs into namespaces, so it is likely that some sort of concatenation of namespace and ID will occur). Also, in section 1.2.4 Modules: Change IDL modules are represented in WSDL as prefixes to the names contained in the module. These prefixes are concatenated using a as a separator. For example, an entit y in a modul e will have the nam e in WSDL. to the following: IDL modules are represented in WSDL as module name prefixes to the names contained in the module. The module prefix to use shall be the pragma prefix associated with to the IDL module definition (or the empty string if there is no pragma prefix directive) concatenated.to the module name, using a as a separator. These module prefixes are concatenated using a as a separator. For example, an entit y in a modul e with pragma prefi x will have the nam e in WSDL. If there were no pragma prefix directive, the same IDL entit y in modul e would have the nam e in WSDL. Fix all examples to append the pragma prefix before the module name (at least pu t before the module name, if there is no pragma prefix in the IDL example). Actions taken: December 17, 2002: received issue -- ---------------------------------------------------- Tom Rutt email: tom@coastin.com; trutt@fsw.fujitsu.com Tel: +1 732 801 5744 Fax: +1 732 774 5133 Date: Wed, 19 Mar 2003 11:58:48 -0500 From: Jishnu Mukerji Organization: Software Global Business Unit, Hewlett-Packard User-Agent: Mozilla/5.0 (Windows; U; Win98; en-US; rv:1.0.2) Gecko/20021120 Netscape/7.01 X-Accept-Language: en-us, en To: tom@coastin.com Cc: "corbawsdl-ftf@omg.org" , iwasa Subject: Re: ISSUE 5785 Proposal In post 3.0 IDL there is no "pragma prefix". In post 3.0 IDL there is only "typeid". Also, pre 3.0 IDL one could use "pragma id" instead of "pragma prefix". Would it be the intention to use the typeid specified by "typeid" or "pragma id" if they are present too? If that is the case, why don't we simply say that the effective IDL Repository ID will be used, with all ":" by "."? In that case the general namespace name will be of the form "IDL.". Jishnu. Tom Rutt wrote: Upon further reflection, I think the proposal should be modified slightly. If there is no pragma prefix associated with the module, the module prefix should be the module name alone (without the ":" prefix). I think there might be problems with a name beginning with the ":" symbol, and also it is unnecessary. Thus the last example in the proposal should be "M.X" instead of ":M.X" Tom Rutt Tom Rutt wrote: Issue 5785: CORBA-WSDL/SOAP: Section 1.2.2.1 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: Section 1.2.2.1 - tns is a single namespace for all mapped IDL from the entire corba world. The module names for IDL across this vast namespace, need to be sufficiently qualified with pre-fixes (such as Type ID, or Pragma Prefix) to avoid module name collisions. Proposed Resolution: FTF needs to determine appropriate prefixes or namespaces to assure module name uniqueness in the generated WSDL Discussion: The brute force approach is to use the scheme that is used for IDL form of repository ID for a Module. Prefix the module name with the string from the pragma prefix IDL directive, usin g (i.e. An OMG IDL pragma of the format: #pragma prefi x ) Proposed Resolution: The WSDL name for the IDL Module should be patterned after the IDL form for the repository ID for the Module. Append the pragma prefix (if present, in front of the module name. If the module definition has no corresponding pragma prefix directive, the prefix string shall be empty. If pragma prefixes are not used, there could be collisions. If such collisions were detected in the mapped WSDL document, they could be resolved by the addition of an appropriate pragma prefix to one of the IDL module definitions. Need to fix all examples to use this pragma prefix in module name prefixes Proposed Revised Text: Remove the following text from th e row of the table in 1.2.2.1 (There are still some issues to be addressed at the FTF stage around how we should map repository IDs into namespaces, so it is likely that some sort of concatenation of namespace and ID will occur). Also, in section 1.2.4 Modules: Change IDL modules are represented in WSDL as prefixes to the names contained in the module. These prefixes are concatenated using a as a separator. For example, an entit y in a modul e will have the nam e in WSDL. to the following: IDL modules are represented in WSDL as module name prefixes to the names contained in the module. The module prefix to use shall be the pragma prefix associated with to the IDL module definition (or the empty string if there is no pragma prefix directive) concatenated.to the module name, using a as a separator. These module prefixes are concatenated using a as a separator. For example, an entit y in a modul e with pragma prefi x will have the nam e in WSDL. If there were no pragma prefix directive, the same IDL entit y in modul e would have the nam e in WSDL. Fix all examples to append the pragma prefix before the module name (at least pu t before the module name, if there is no pragma prefix in the IDL example). Actions taken: December 17, 2002: received issue -- Jishnu Mukerji Senior Systems Architect 1001 Frontier Road, Suite 300 Technology Office Bridgewater NJ 08807, USA Software Global Business Unit Tel: +1 908 243 8924 Hewlett-Packard Company Fax: +1 908 243 8850 mailto: jishnu@hp.com Date: Tue, 18 Mar 2003 20:16:13 -0500 From: Tom Rutt Reply-To: tom@coastin.com User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.0.2) Gecko/20021120 Netscape/7.01 X-Accept-Language: en-us, en To: corbawsdl-ftf@omg.org CC: tom@coastin.com Subject: Proposed Issue Resolutions for corbawsdl FTF X-OriginalArrivalTime: 19 Mar 2003 01:19:15.0421 (UTC) FILETIME=[8E6DA8D0:01C2EDB5] Well, the Business committee has not yet adopted the CORBA WSDL mapping spec. Thus we will have to extend the due dated for the FTF to be after the Paris meeting. I have provided proposed resolutions for all the current issues on the list. Please provide any comments on these proposed resolutions on the ftf list. After a comments round, we can then proceed to votes on the resolutions. Tom Rutt corbawsdl ftf Chair. The proposed changes are attached to this mail as html. To get them in the archive I also will send a separate email for each issue. Please respond to the separate emails per issue, so the discussion will get into the Issues database for each issue. -- ---------------------------------------------------- Tom Rutt email: tom@coastin.com; trutt@fsw.fujitsu.com Tel: +1 732 801 5744 Fax: +1 732 774 5133 xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:st1="urn:schemas-microsoft-com:office:smarttags" xmlns="http://www.w3.org/TR/REC-html40"> Issue 5785: CORBA-WSDL/SOAP: Section 1.2.2.1 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: Section 1.2.2.1 - tns is a single namespace for all mapped IDL from the entire corba world. The module names for IDL across this vast namespace, need to be sufficiently qualified with pre-fixes (such as Type ID, or Pragma Prefix) to avoid module name collisions. Proposed Resolution: FTF needs to determine appropriate prefixes or namespaces to assure module name uniqueness in the generated WSDL Discussion: The brute force approach is to use the scheme that is used for IDL form of repository ID for a Module. Prefix the module name with the string from the pragma prefix IDL directive, usin g (i.e. An OMG IDL pragma of the format: #pragma prefi x ) Proposed Resolution: The WSDL name for the IDL Module should be patterned after the IDL form for the repository ID for the Module. Append the pragma prefix (if present, in front of the module name. If the module definition has no corresponding pragma prefix directive, the prefix string shall be empty. If pragma prefixes are not used, there could be collisions. If such collisions were detected in the mapped WSDL document, they could be resolved by the addition of an appropriate pragma prefix to one of the IDL module definitions. Need to fix all examples to use this pragma prefix in module name prefixes Proposed Revised Text: Remove the following text from th e row of the table in 1.2.2.1 (There are still some issues to be addressed at the FTF stage around how we should map repository IDs into namespaces, so it is likely that some sort of concatenation of namespace and ID will occur). Also, in section 1.2.4 Modules: Change IDL modules are represented in WSDL as prefixes to the names contained in the module. These prefixes are concatenated using a as a separator. For example, an entit y in a modul e will have the nam e in WSDL. to the following: IDL modules are represented in WSDL as module name prefixes to the names contained in the module. The module prefix to use shall be the pragma prefix associated with to the IDL module definition (or the empty string if there is no pragma prefix directive) concatenated.to the module name, using a as a separator. These module prefixes are concatenated using a as a separator. For example, an entit y in a modul e with pragma prefi x will have the nam e in WSDL. If there were no pragma prefix directive, the same IDL entit y in modul e would have the nam e in WSDL. Fix all examples to append the pragma prefix before the module name (at least pu t before the module name, if there is no pragma prefix in the IDL example). Actions taken: December 17, 2002: received issue xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:st1="urn:schemas-microsoft-com:office:smarttags" xmlns="http://www.w3.org/TR/REC-html40"> Issue 5786: CORBA-WSDL/SOAP: Section 1.2.3 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: Section 1.2.3 - need to have appropriate text to describe the nature of the source "hint". Examples might be included, using filenames. Proposed Resolution: FTF should determine appropriate text. Discussion: The nature of the hint can remain an implementation specific matter. Resolution: Close issue with no revision. Revised Text: None Actions taken: December 17, 2002: received issue xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:st1="urn:schemas-microsoft-com:office:smarttags" xmlns="http://www.w3.org/TR/REC-html40"> Issue 5787: Section 1.2.3 repository id inconsistency (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: Section 1.2.3 - there is an inconsistency, in that the documentation schema for the repository ID does not include the version of the mapping, as included in the URL source documentation schema. Knowing what version of the IDL to WSDL mapping produced the WSDL is important for gateway design. Proposed Resolution: add mapping version to the respositoryID documentation schema. Discussion: This is a bug. There is also two typographical errors in the schema for IDL source Proposed Resolution: Add mapping version to the repositoryID documentation schema Proposed Revised Text: In section 1.2.3, schema for IDLsource: change: IDL/WSDL Mapping Info to IDL/WSDL Mapping Info also , in the same schema change: type="xsd:string" minOccurs="1" maxOccurs="1"/> to also, in the example text after the first schema, change (to match the schema name for the complex type): IDL Source 1.0 to: IDL Source 1.0 also, change the following text for the repositoryID form schema and example from: The generated WSDL will then use this schema to indicate the mapping information as shown in the following example: repository-id to the following: IDL Mapped Repository ID The generated WSDL will then use this schema to indicate the mapping information as shown in the following example: repositoryIDtext 1.0 Actions taken: December 17, 2002: received issue xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:st1="urn:schemas-microsoft-com:office:smarttags" xmlns="http://www.w3.org/TR/REC-html40"> Issue 5788: CORBA-WSDL/SOAP: section 1.2.4 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: section 1.2.4 - The sentence "The most significant is that XML flattens the constructed namespace to a single targeted namespace when an XML schema is imported." does not seem quite correct. Avoiding a large number of files for the wsdl processor to deal with is a major reason cited by the submitters. Proposed Solution: Replace the offending sentence, with: "Having a separate namespace for each imported module results in a large number of files for the WSDL processor to deal with when constructing a gateway." Proposed Resolution: Agree to editorial fix. Proposed Revised Text: In section 1.2.4 change: The most significant is that XML flattens the constructed namespace to a single targeted namespace when an XML schema is imported. to: Having a separate namespace for each imported module results in a large number of files for the WSDL processor to deal with when constructing a gateway. Actions taken: December 17, 2002: received issue xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:st1="urn:schemas-microsoft-com:office:smarttags" xmlns="http://www.w3.org/TR/REC-html40"> Issue 5790: CORBA-WSDL/SOAP: Section 1.2.7.10 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: Section 1.2.7.10 - Only public state members are mapped. Since use of the Dynamic Any api allows access to both the private and public members, this restriction was challenged by the AB revieweras unwarranted. Also the sentence "For example, DII/DSI bridges will be unable to process value types." is not justified. The dynamic ANY api can be used to access value type state members in a DII/DSI based gateway. If private state cannot be represented, an input value type parameter which has private state will not be mapped to the IDL server. Proposed Solution: Delete offending sentence about dynamic gateways. Unless the private member restriction can be justified, change the spec to map both private and public value type state members. Discussion: Since the dynamic ANY api can be used to access all value type state members in a DII/DSI gateway, the restriction to not mapping private state members needs to be removed. Proposed Resolution: Remove the restriction of not mapping Private state members. Proposed Revised Text: In 1.2.7.10: Remove the following text from the first sentence: Change: Basic valuetypes are mapped to structs, as noted above. Private fields are ignored. For example, consider the following: // IDL valuetype sampleX { public short a; private long b; } The private fiel d is ignored, so this maps to: to the following: Basic valuetypes are mapped to structs, as noted above. Both Public and Private fields are mapped. For example, consider the following: // IDL valuetype sampleX { public short a; private long b; } This maps to: Actions taken: December 18, 2002: received issue xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:st1="urn:schemas-microsoft-com:office:smarttags" xmlns="http://www.w3.org/TR/REC-html40"> Issue 5791: CORBA-WSDL/SOAP: Section 1.2.7.10.3 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: Section 1.2.7.10.3 - need to represent value type in IDL sequence as an IdRef/value choice. Proposed Resolution: fix spec to use the idRef/Value choice for sequence. Discussion: Agree to bug fix. Also need resolve editorial problems pointed out in bold by the reviewer in ptc/3-01-14. Proposed Resolution: include IDL sequence including a value type in the mapping algorithm generation of the IdRef/value choice type. Proposed Revised Text: In 1.2.7.10 change: Mapping of value types as struct or value type members IDL value types can have recursive definitions, allowing complex structures, such as graphs, to be represented. This occurs whenever a value type has a state member that is the same type as its containing value type. Such cycles in an instance graph can b e though a series of value types, which contain member Due to its complexity, support at run time for mapping shared value types is a separate, optional, conformance point for this specification. Whenever an IDL struct or an IDL value type includes a member that is a value type, the corresponding sequence element for the value type member must be mapped to an xsd:choice element, containing either: an element which is an that valuetype or, [Reviewer: Please clarify] an element which is a reference to a value type instance included in the same XML document. to the following: Mapping of value types as struct or value type members IDL value types can have recursive definitions, allowing complex structures, such as graphs, to be represented. This occurs whenever a value type has a state member that is the same type as its containing value type. Such cycles in an instance graph can be s of eac h through a series of value types, which contain member Due to its complexity, support at run time for mapping shared value types is a separate, optional, conformance point for this specification. Whenever an IDL struct, sequence, or value type includes a member that is a value type, the corresponding sequence element for the value type member must be mapped to an xsd:choice element, containing either: an element which is a value of that valuetype or, an element which is a reference to a value type instance included in the same XML document. Actions taken: December 17, 2002: received issue xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:st1="urn:schemas-microsoft-com:office:smarttags" xmlns="http://www.w3.org/TR/REC-html40"> Issue 5792: CORBA-WSDL/SOAP: section 1.2.8.2 , 1.2.8.4 and 1.2.8.5 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: section 1.2.8.2 , 1.2.8.4 and 1.2.8.5 - In all the Example WSDL, the target namespace in definitions and the namespace in import using URL "http://tempuri.org/anExample/.." do not seem correct. Proposed Resolution: fix the example after resolving the Type prefix FTF issue on namespace and module name uniqueness. Proposed Resolution: Fix the examples to use the correct target namespace url. Proposed Revised Text: In 1.2.8.2, 1.2.8.4, and 1.2.8.4, change: targetNamespace="http://tempuri.org/anExample/definitions" xmlns:tns="http://tempuri.org/anExample/definitions" xmlns:xsd1="http://tempuri.org/anExample/schemas" xmlns:soap="http://schemas.xmlsoap.org/wsdl/soap/" xmlns="http://schemas.xmlsos of each otherother Date: Wed, 02 Apr 2003 11:35:32 -0500 From: Tom Rutt Reply-To: tom@coastin.com User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.0.2) Gecko/20021120 Netscape/7.01 X-Accept-Language: en-us, en To: Jishnu Mukerji CC: "corbawsdl-ftf@omg.org" , iwasa Subject: Re: ISSUE 5785 Proposal We need to get working hard on resolving this one. Jishnu, could you please give a reason why we need the "IDL" prefix, if we agree to use the IDL form for respositoy ID? Why cannot it be implied. Another possibility would be to only add the prefix when a module name collision occurs in the WSDL document. Add the prefix to the first module which has the same name as a module which has already been included in the WSDL document. What does everyone think on this. Tom Rutt Jishnu Mukerji wrote: In post 3.0 IDL there is no "pragma prefix". In post 3.0 IDL there is only "typeid". Also, pre 3.0 IDL one could use "pragma id" instead of "pragma prefix". Would it be the intention to use the typeid specified by "typeid" or "pragma id" if they are present too? If that is the case, why don't we simply say that the effective IDL Repository ID will be used, with all ":" by "."? In that case the general namespace name will be of the form "IDL.". Jishnu. Tom Rutt wrote: Upon further reflection, I think the proposal should be modified slightly. If there is no pragma prefix associated with the module, the module prefix should be the module name alone (without the ":" prefix). I think there might be problems with a name beginning with the ":" symbol, and also it is unnecessary. Thus the last example in the proposal should be "M.X" instead of ":M.X" Tom Rutt Tom Rutt wrote: Issue 5785: CORBA-WSDL/SOAP: Section 1.2.2.1 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: Section 1.2.2.1 - tns is a single namespace for all mapped IDL from the entire corba world. The module names for IDL across this vast namespace, need to be sufficiently qualified with pre-fixes (such as Type ID, or Pragma Prefix) to avoid module name collisions. Proposed Resolution: FTF needs to determine appropriate prefixes or namespaces to assure module name uniqueness in the generated WSDL Discussion: The brute force approach is to use the scheme that is used for IDL form of repository ID for a Module. Prefix the module name with the string from the pragma prefix IDL directive, usin g (i.e. An OMG IDL pragma of the format: #pragma prefi x ) Proposed Resolution: The WSDL name for the IDL Module should be patterned after the IDL form for the repository ID for the Module. Append the pragma prefix (if present, in front of the module name. If the module definition has no corresponding pragma prefix directive, the prefix string shall be empty. If pragma prefixes are not used, there could be collisions. If such collisions were detected in the mapped WSDL document, they could be resolved by the addition of an appropriate pragma prefix to one of the IDL module definitions. Need to fix all examples to use this pragma prefix in module name prefixes Proposed Revised Text: Remove the following text from th e row of the table in 1.2.2.1 (There are still some issues to be addressed at the FTF stage around how we should map repository IDs into namespaces, so it is likely that some sort of concatenation of namespace and ID will occur). Also, in section 1.2.4 Modules: Change IDL modules are represented in WSDL as prefixes to the names contained in the module. These prefixes are concatenated using a as a separator. For example, an entit y in a modul e will have the nam e in WSDL. to the following: IDL modules are represented in WSDL as module name prefixes to the names contained in the module. The module prefix to use shall be the pragma prefix associated with to the IDL module definition (or the empty string if there is no pragma prefix directive) concatenated.to the module name, using a as a separator. These module prefixes are concatenated using a as a separator. For example, an entit y in a modul e with pragma prefi x will have the nam e in WSDL. If there were no pragma prefix directive, the same IDL entit y in modul e would have the nam e in WSDL. Fix all examples to append the pragma prefix before the module name (at least pu t before the module name, if there is no pragma prefix in the IDL example). Actions taken: December 17, 2002: received issue -- ---------------------------------------------------- Tom Rutt email: tom@coastin.com; trutt@fsw.fujitsu.com Tel: +1 732 801 5744 Fax: +1 732 774 5133 Date: Wed, 02 Apr 2003 13:28:54 -0500 From: Jishnu Mukerji Organization: Software Global Business Unit, Hewlett-Packard User-Agent: Mozilla/5.0 (Windows; U; Win98; en-US; rv:1.0.2) Gecko/20021120 Netscape/7.01 X-Accept-Language: en-us, en To: tom@coastin.com Cc: "corbawsdl-ftf@omg.org" , iwasa Subject: Re: ISSUE 5785 Proposal Tom Rutt wrote: We need to get working hard on resolving this one. Jishnu, could you please give a reason why we need the "IDL" prefix, if we agree to use the IDL form for respositoy ID? Why cannot it be implied. I don't think it is important to include the "IDL:" prefix What is important though is to cover all possible combination of typeprefix and typeid directives. One complication arises because the only way to create a IDL type Rep id with a new version number using these directives is to use the "typeid" directive and in it specify the entire Rep Id including the "IDL:". Also, if there is a typeid directive for a scoped name, then for that scoped name any typeprefix directive that may be active does not apply. This precedence rule is the same as for pragma prefix and pragma id. So here is how we can cover all bases: 1. If there are no pragma or typeid/typeprefix directives that applies to a module name then the scoped name of the module starting from the root module name using the "." character as separator between names shall be the effective Id used in the corresponding WSDL. 2. If there is a pragma prefix or a typeprefix directive that applies to the module name then the prefix specified in the directive/pragma shall be of the form: ":" If no such directives apply to the module name in question then the effective Id shall be the same as in (1). 3. If there is a pragma version that applies to the scopename then the version string shall be appened to the effective Id obtained in (2), that is the effective Id will be of the form: ":" 4. If a pragma id or a typeid directive applies to the module name in question then: 4a. If the first four characters of the string specified in the directive/pragma is "IDL:" then the string with the "IDL:" prefix removed and with the "/" characters in the remaining string substituted by the "." character shall be the effective Id. 4b. If the Repository Id type prefix in the directive/pragma is something other than "IDL:" then ......... 4b choice 1 "the entire id string shall be used as the effective Id" 4b choice 2 "it shall be flagged as an error" 4b choice 3 "the id shall be ignored and the effective Id as computed in (1) shall be used. I think choice 1 (with some munging rule to escape out disallowed characters) is preferable since it allows the use of domain specific Ids. The only problem is that there is the infinitisimal chance of a id clash in certain situations. Comments? Jishnu. Proposed Resolution: FTF needs to determine appropriate prefixes or namespaces to assure module name uniqueness in the generated WSDL Discussion: The brute force approach is to use the scheme that is used for IDL form of repository ID for a Module. Prefix the module name with the string from the pragma prefix IDL directive, usin g (i.e. An OMG IDL pragma of the format: #pragma prefi x ) Proposed Resolution: The WSDL name for the IDL Module should be patterned after the IDL form for the repository ID for the Module. Append the pragma prefix (if present, in front of the module name. If the module definition has no corresponding pragma prefix directive, the prefix string shall be empty. Suggest that the rule be stated explicitly as I propose above, and the consequent changes be made to the proposed new text. If pragma prefixes are not used, there could be collisions. If such collisions were detected in the mapped WSDL document, they could be resolved by the addition of an appropriate pragma prefix to one of the IDL module definitions. What would the appropriate pargma prefix be? Do we need to nail it down a bit more or leave it open? Need to fix all examples to use this pragma prefix in module name prefixes Proposed Revised Text: Remove the following text from th e row of the table in 1.2.2.1 (There are still some issues to be addressed at the FTF stage around how we should map repository IDs into namespaces, so it is likely that some sort of concatenation of namespace and ID will occur). Also, in section 1.2.4 Modules: Change IDL modules are represented in WSDL as prefixes to the names contained in the module. These prefixes are concatenated using a as a separator. For example, an entit y in a modul e will have the nam e in WSDL. to the following: IDL modules are represented in WSDL as module name prefixes to the names contained in the module. The module prefix to use shall be the pragma prefix associated with to the IDL module definition (or the empty string if there is no pragma prefix directive) concatenated.to the module name, using a as a separator. These module prefixes are concatenated using a as a separator. For example, an entit y in a modul e with pragma prefi x will have the nam e in WSDL. If there were no pragma prefix directive, the same IDL entit y in modul e would have the nam e in WSDL. Fix all examples to append the pragma prefix before the module name (at least pu t before the module name, if there is no pragma prefix in the IDL example). Actions taken: December 17, 2002: received issue Date: Thu, 03 Apr 2003 13:20:23 -0500 From: Tom Rutt Reply-To: tom@coastin.com User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.0.2) Gecko/20021120 Netscape/7.01 X-Accept-Language: en-us, en To: Jishnu Mukerji CC: "corbawsdl-ftf@omg.org" , iwasa Subject: Re: ISSUE 5785 Proposal What about the idea of only adding the prefix when a collision is detected in a single WSDL document? Most of the time there will be no collision, thus the names would be quite shorter. I realize that the prefix might be added to a module in some wsdl docs while in others (depending which module does includes from which other module) that same mopule might not have a prefix. However, the schema for each wsdl document will be consistent and without collision. Does it matter that some WSDL documents will use a prefix for a translated IDL module while others will not? Tom Rutt Jishnu Mukerji wrote: Tom Rutt wrote: We need to get working hard on resolving this one. Jishnu, could you please give a reason why we need the "IDL" prefix, if we agree to use the IDL form for respositoy ID? Why cannot it be implied. I don't think it is important to include the "IDL:" prefix What is important though is to cover all possible combination of typeprefix and typeid directives. One complication arises because the only way to create a IDL type Rep id with a new version number using these directives is to use the "typeid" directive and in it specify the entire Rep Id including the "IDL:". Also, if there is a typeid directive for a scoped name, then for that scoped name any typeprefix directive that may be active does not apply. This precedence rule is the same as for pragma prefix and pragma id. So here is how we can cover all bases: 1. If there are no pragma or typeid/typeprefix directives that applies to a module name then the scoped name of the module starting from the root module name using the "." character as separator between names shall be the effective Id used in the corresponding WSDL. 2. If there is a pragma prefix or a typeprefix directive that applies to the module name then the prefix specified in the directive/pragma shall be of the form: ":" If no such directives apply to the module name in question then the effective Id shall be the same as in (1). 3. If there is a pragma version that applies to the scopename then the version string shall be appened to the effective Id obtained in (2), that is the effective Id will be of the form: ":" 4. If a pragma id or a typeid directive applies to the module name in question then: 4a. If the first four characters of the string specified in the directive/pragma is "IDL:" then the string with the "IDL:" prefix removed and with the "/" characters in the remaining string substituted by the "." character shall be the effective Id. 4b. If the Repository Id type prefix in the directive/pragma is something other than "IDL:" then ......... 4b choice 1 "the entire id string shall be used as the effective Id" 4b choice 2 "it shall be flagged as an error" 4b choice 3 "the id shall be ignored and the effective Id as computed in (1) shall be used. I think choice 1 (with some munging rule to escape out disallowed characters) is preferable since it allows the use of domain specific Ids. The only problem is that there is the infinitisimal chance of a id clash in certain situations. Comments? Jishnu. Proposed Resolution: FTF needs to determine appropriate prefixes or namespaces to assure module name uniqueness in the generated WSDL Discussion: The brute force approach is to use the scheme that is used for IDL form of repository ID for a Module. Prefix the module name with the string from the pragma prefix IDL directive, usin g (i.e. An OMG IDL pragma of the format: #pragma prefi x ) Proposed Resolution: The WSDL name for the IDL Module should be patterned after the IDL form for the repository ID for the Module. Append the pragma prefix (if present, in front of the module name. If the module definition has no corresponding pragma prefix directive, the prefix string shall be empty. Suggest that the rule be stated explicitly as I propose above, and the consequent changes be made to the proposed new text. If pragma prefixes are not used, there could be collisions. If such collisions were detected in the mapped WSDL document, they could be resolved by the addition of an appropriate pragma prefix to one of the IDL module definitions. What would the appropriate pargma prefix be? Do we need to nail it down a bit more or leave it open? Need to fix all examples to use this pragma prefix in module name prefixes Proposed Revised Text: Remove the following text from th e row of the table in 1.2.2.1 (There are still some issues to be addressed at the FTF stage around how we should map repository IDs into namespaces, so it is likely that some sort of concatenation of namespace and ID will occur). Also, in section 1.2.4 Modules: Change IDL modules are represented in WSDL as prefixes to the names contained in the module. These prefixes are concatenated using a as a separator. For example, an entit y in a modul e will have the nam e in WSDL. to the following: IDL modules are represented in WSDL as module name prefixes to the names contained in the module. The module prefix to use shall be the pragma prefix associated with to the IDL module definition (or the empty string if there is no pragma prefix directive) concatenated.to the module name, using a as a separator. These module prefixes are concatenated using a as a separator. For example, an entit y in a modul e with pragma prefi x will have the nam e in WSDL. If there were no pragma prefix directive, the same IDL entit y in modul e would have the nam e in WSDL. Fix all examples to append the pragma prefix before the module name (at least pu t before the module name, if there is no pragma prefix in the IDL example). Actions taken: December 17, 2002: received issue -- ---------------------------------------------------- Tom Rutt email: tom@coastin.com; trutt@fsw.fujitsu.com Tel: +1 732 801 5744 Fax: +1 732 774 5133 Date: Thu, 03 Apr 2003 15:43:08 -0500 From: Jishnu Mukerji Organization: Software Global Business Unit, Hewlett-Packard User-Agent: Mozilla/5.0 (Windows; U; Win98; en-US; rv:1.0.2) Gecko/20021120 Netscape/7.01 X-Accept-Language: en-us, en To: tom@coastin.com Cc: "corbawsdl-ftf@omg.org" , iwasa Subject: Re: ISSUE 5785 Proposal Tom Rutt wrote: What about the idea of only adding the prefix when a collision is detected in a single WSDL document? Most of the time there will be no collision, thus the names would be quite shorter. I realize that the prefix might be added to a module in some wsdl docs while in others (depending which module does includes from which other module) that same mopule might not have a prefix. However, the schema for each wsdl document will be consistent and without collision. Does it matter that some WSDL documents will use a prefix for a translated IDL module while others will not? Whatever is it that we do should be such that the info can be used to be able to create a consistent IDL file in the reverse mapping. I don't know that there is any requirement other than that and that the namespace in the WSDL document should be unique. Jishnu. Date: Fri, 04 Apr 2003 12:17:08 -0500 From: Tom Rutt Reply-To: tom@coastin.com User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.0.2) Gecko/20021120 Netscape/7.01 X-Accept-Language: en-us, en To: Jishnu Mukerji CC: "corbawsdl-ftf@omg.org" , iwasa Subject: Re: ISSUE 5785 Proposal We are using a single name space for all wsdl documents which come from IDL definitions. However, the WSDL processor is only processing one WSDL document for each translated IDL specification. All the included modules in the original IDL "file " which is the source are physically included in the WSDL document. Thus some wsdl documents could, given my proposal of prefixing only when collision occurs in a single wsdl document, have different identifiers for IDL types than another document which has those same types included thru a different IDL include sequence. I am not sure this is a problem however. Tom Rutt Jishnu Mukerji wrote: Tom Rutt wrote: What about the idea of only adding the prefix when a collision is detected in a single WSDL document? Most of the time there will be no collision, thus the names would be quite shorter. I realize that the prefix might be added to a module in some wsdl docs while in others (depending which module does includes from which other module) that same mopule might not have a prefix. However, the schema for each wsdl document will be consistent and without collision. Does it matter that some WSDL documents will use a prefix for a translated IDL module while others will not? Whatever is it that we do should be such that the info can be used to be able to create a consistent IDL file in the reverse mapping. I don't know that there is any requirement other than that and that the namespace in the WSDL document should be unique. Jishnu. -- ---------------------------------------------------- Tom Rutt email: tom@coastin.com; trutt@fsw.fujitsu.com Tel: +1 732 801 5744 Fax: +1 732 774 5133 Date: Fri, 04 Apr 2003 12:27:22 -0500 From: Tom Rutt Reply-To: tom@coastin.com User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.0.2) Gecko/20021120 Netscape/7.01 X-Accept-Language: en-us, en To: tom@coastin.com CC: Jishnu Mukerji , "corbawsdl-ftf@omg.org" , iwasa Subject: Re: ISSUE 5785 Proposal and the reverse mapping By the way, we have to decide what to map to IDL Module in the reverse WSDL to IDL mapping. The last submission maps "service" to module, but that is not correct since service is for an instance of a collection of ports. We need to map a "type" thing to IDL module. Tom Rutt Tom Rutt wrote: We are using a single name space for all wsdl documents which come from IDL definitions. However, the WSDL processor is only processing one WSDL document for each translated IDL specification. All the included modules in the original IDL "file " which is the source are physically included in the WSDL document. Thus some wsdl documents could, given my proposal of prefixing only when collision occurs in a single wsdl document, have different identifiers for IDL types than another document which has those same types included thru a different IDL include sequence. I am not sure this is a problem however. Tom Rutt Jishnu Mukerji wrote: Tom Rutt wrote: What about the idea of only adding the prefix when a collision is detected in a single WSDL document? Most of the time there will be no collision, thus the names would be quite shorter. I realize that the prefix might be added to a module in some wsdl docs while in others (depending which module does includes from which other module) that same mopule might not have a prefix. However, the schema for each wsdl document will be consistent and without collision. Does it matter that some WSDL documents will use a prefix for a translated IDL module while others will not? Whatever is it that we do should be such that the info can be used to be able to create a consistent IDL file in the reverse mapping. I don't know that there is any requirement other than that and that the namespace in the WSDL document should be unique. Jishnu. -- ---------------------------------------------------- Tom Rutt email: tom@coastin.com; trutt@fsw.fujitsu.com Tel: +1 732 801 5744 Fax: +1 732 774 5133 Date: Wed, 02 Apr 2003 11:45:26 -0500 From: Tom Rutt Reply-To: tom@coastin.com User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.0.2) Gecko/20021120 Netscape/7.01 X-Accept-Language: en-us, en To: "corbawsdl-ftf@omg.org" , iwasa Subject: Vote impending for non controversial corbawsdl ftf issues So far the only issue which has had any discussion is ISSUE 5785. The issue on examples (5792) depends on this issue resolution as well. Thus I will send out a vote with the other proposed resolutions on this coming Monday. Please look at the attached resolutions for wordsmithing before the vote is sent out. Please provided your comments on each proposal in a separate mail with the issue number in the title so it will get into the issues email archive. TomRutt FTF chair -- ---------------------------------------------------- Tom Rutt email: tom@coastin.com; trutt@fsw.fujitsu.com Tel: +1 732 801 5744 Fax: +1 732 774 5133 xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:st1="urn:schemas-microsoft-com:office:smarttags" xmlns="http://www.w3.org/TR/REC-html40"> Issue 5785: CORBA-WSDL/SOAP: Section 1.2.2.1 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: Section 1.2.2.1 - tns is a single namespace for all mapped IDL from the entire corba world. The module names for IDL across this vast namespace, need to be sufficiently qualified with pre-fixes (such as Type ID, or Pragma Prefix) to avoid module name collisions. Proposed Resolution: FTF needs to determine appropriate prefixes or namespaces to assure module name uniqueness in the generated WSDL Discussion: The brute force approach is to use the scheme that is used for IDL form of repository ID for a Module. Prefix the module name with the string from the pragma prefix IDL directive, usin g (i.e. An OMG IDL pragma of the format: #pragma prefi x ) Proposed Resolution: The WSDL name for the IDL Module should be patterned after the IDL form for the repository ID for the Module. Append the pragma prefix (if present, in front of the module name. If the module definition has no corresponding pragma prefix directive, the prefix string shall be empty. If pragma prefixes are not used, there could be collisions. If such collisions were detected in the mapped WSDL document, they could be resolved by the addition of an appropriate pragma prefix to one of the IDL module definitions. Need to fix all examples to use this pragma prefix in module name prefixes Proposed Revised Text: Remove the following text from th e row of the table in 1.2.2.1 (There are still some issues to be addressed at the FTF stage around how we should map repository IDs into namespaces, so it is likely that some sort of concatenation of namespace and ID will occur). Also, in section 1.2.4 Modules: Change IDL modules are represented in WSDL as prefixes to the names contained in the module. These prefixes are concatenated using a as a separator. For example, an entit y in a modul e will have the nam e in WSDL. to the following: IDL modules are represented in WSDL as module name prefixes to the names contained in the module. The module prefix to use shall be the pragma prefix associated with to the IDL module definition (or the empty string if there is no pragma prefix directive) concatenated.to the module name, using a as a separator. These module prefixes are concatenated using a as a separator. For example, an entit y in a modul e with pragma prefi x will have the nam e in WSDL. If there were no pragma prefix directive, the same IDL entit y in modul e would have the nam e in WSDL. Fix all examples to append the pragma prefix before the module name (at least pu t before the module name, if there is no pragma prefix in the IDL example). Actions taken: December 17, 2002: received issue xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:st1="urn:schemas-microsoft-com:office:smarttags" xmlns="http://www.w3.org/TR/REC-html40"> Issue 5786: CORBA-WSDL/SOAP: Section 1.2.3 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: Section 1.2.3 - need to have appropriate text to describe the nature of the source "hint". Examples might be included, using filenames. Proposed Resolution: FTF should determine appropriate text. Discussion: The nature of the hint can remain an implementation specific matter. Resolution: Close issue with no revision. Revised Text: None Actions taken: December 17, 2002: received issue xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:st1="urn:schemas-microsoft-com:office:smarttags" xmlns="http://www.w3.org/TR/REC-html40"> Issue 5787: Section 1.2.3 repository id inconsistency (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: Section 1.2.3 - there is an inconsistency, in that the documentation schema for the repository ID does not include the version of the mapping, as included in the URL source documentation schema. Knowing what version of the IDL to WSDL mapping produced the WSDL is important for gateway design. Proposed Resolution: add mapping version to the respositoryID documentation schema. Discussion: This is a bug. There is also two typographical errors in the schema for IDL source Proposed Resolution: Add mapping version to the repositoryID documentation schema Proposed Revised Text: In section 1.2.3, schema for IDLsource: change: IDL/WSDL Mapping Info to IDL/WSDL Mapping Info also , in the same schema change: type="xsd:string" minOccurs="1" maxOccurs="1"/> to also, in the example text after the first schema, change (to match the schema name for the complex type): IDL Source 1.0 to: IDL Source 1.0 also, change the following text for the repositoryID form schema and example from: The generated WSDL will then use this schema to indicate the mapping information as shown in the following example: repository-id to the following: IDL Mapped Repository ID The generated WSDL will then use this schema to indicate the mapping information as shown in the following example: repositoryIDtext 1.0 Actions taken: December 17, 2002: received issue xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:st1="urn:schemas-microsoft-com:office:smarttags" xmlns="http://www.w3.org/TR/REC-html40"> Issue 5788: CORBA-WSDL/SOAP: section 1.2.4 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: section 1.2.4 - The sentence "The most significant is that XML flattens the constructed namespace to a single targeted namespace when an XML schema is imported." does not seem quite correct. Avoiding a large number of files for the wsdl processor to deal with is a major reason cited by the submitters. Proposed Solution: Replace the offending sentence, with: "Having a separate namespace for each imported module results in a large number of files for the WSDL processor to deal with when constructing a gateway." Proposed Resolution: Agree to editorial fix. Proposed Revised Text: In section 1.2.4 change: The most significant is that XML flattens the constructed namespace to a single targeted namespace when an XML schema is imported. to: Having a separate namespace for each imported module results in a large number of files for the WSDL processor to deal with when constructing a gateway. Actions taken: December 17, 2002: received issue xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:st1="urn:schemas-microsoft-com:office:smarttags" xmlns="http://www.w3.org/TR/REC-html40"> Issue 5790: CORBA-WSDL/SOAP: Section 1.2.7.10 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: Section 1.2.7.10 - Only public state members are mapped. Since use of the Dynamic Any api allows access to both the private and public members, this restriction was challenged by the AB revieweras unwarranted. Also the sentence "For example, DII/DSI bridges will be unable to process value types." is not justified. The dynamic ANY api can be used to access value type state members in a DII/DSI based gateway. If private state cannot be represented, an input value type parameter which has private state will not be mapped to the IDL server. Proposed Solution: Delete offending sentence about dynamic gateways. Unless the private member restriction can be justified, change the spec to map both private and public value type state members. Discussion: Since the dynamic ANY api can be used to access all value type state members in a DII/DSI gateway, the restriction to not mapping private state members needs to be removed. Proposed Resolution: Remove the restriction of not mapping Private state members. Proposed Revised Text: In 1.2.7.10: Remove the following text from the first sentence: Change: Basic valuetypes are mapped to structs, as noted above. Private fields are ignored. For example, consider the following: // IDL valuetype sampleX { public short a; private long b; } The private fiel d is ignored, so this maps to: to the following: Basic valuetypes are mapped to structs, as noted above. Both Public and Private fields are mapped. For example, consider the following: // IDL valuetype sampleX { public short a; private long b; } This maps to: Actions taken: December 18, 2002: received issue xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:st1="urn:schemas-microsoft-com:office:smarttags" xmlns="http://www.w3.org/TR/REC-html40"> Issue 5791: CORBA-WSDL/SOAP: Section 1.2.7.10.3 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: Section 1.2.7.10.3 - need to represent value type in IDL sequence as an IdRef/value choice. Proposed Resolution: fix spec to use the idRef/Value choice for sequence. Discussion: Agree to bug fix. Also need resolve editorial problems pointed out in bold by the reviewer in ptc/3-01-14. Proposed Resolution: include IDL sequence including a value type in the mapping algorithm generation of the IdRef/value choice type. Proposed Revised Text: In 1.2.7.10 change: Mapping of value types as struct or value type members IDL value types can have recursive definitions, allowing complex structures, such as graphs, to be represented. This occurs whenever a value type has a state member that is the same type as its containing value type. Such cycles in an instance graph can b e though a series of value types, which contain member Due to its complexity, support at run time for mapping shared value types is a separate, optional, conformance point for this specification. Whenever an IDL struct or an IDL value type includes a member that is a value type, the corresponding sequence element for the value type member must be mapped to an xsd:choice element, containing either: an element which is an that valuetype or, [Reviewer: Please clarify] an element which is a reference to a value type instance included in the same XML document. to the following: Mapping of value types as struct or value type members IDL value types can have recursive definitions, allowing complex structures, such as graphs, to be represented. This occurs whenever a value type has a state member that is the same type as its containing value type. Such cycles in an instance graph can be s of eac h through a series of value types, which contain member Due to its complexity, support at run time for mapping shared value types is a separate, optional, conformance point for this specification. Whenever an IDL struct, sequence, or value type includes a member that is a value type, the corresponding sequence element for the value type member must be mapped to an xsd:choice element, containing either: an element which is a value of that valuetype or, an element which is a reference to a value type instance included in the same XML document. Actions taken: December 17, 2002: received issue xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:st1="urn:schemas-microsoft-com:office:smarttags" xmlns="http://www.w3.org/TR/REC-html40"> Issue 5792: CORBA-WSDL/SOAP: section 1.2.8.2 , 1.2.8.4 and 1.2.8.5 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: section 1.2.8.2 , 1.2.8.4 and 1.2.8.5 - In all the Example WSDL, the target namespace in definitions and the namespace in import using URL "http://tempuri.org/anExample/.." do not seem correct. Proposed Resolution: fix the example after resolving the Type prefix FTF issue on namespace and module name uniqueness. Proposed Resolution: Fix the examples to use the correct target namespace url. Proposed Revised Text: In 1.2.8.2, 1.2.8.4, and 1.2.8.4, change: targetNamespace="http://tempuri.org/anExample/definitions" xmlns:tns="http://tempuri.org/anExample/definitions" xmlns:xsd1="http://tempuri.org/anExample/schemas" xmlns:soap="http://schemas.xmlsoap.org/wsdl/soap/" xmlns="http://schemas.xmlsoap.org/wsdl" > to: targetNamespace=" http://www.omg.org/IDL-Mapped/" xmlns:tns="http://www.omg.org/IDL-Mapped/" xmlns:CORBA=http://www.omg.org/IDL-WSDL/1.0/ xmlns:soap="http://schemas.xmlsoap.org/wsdl/soap/" xmlns:wsdl="http://schemas.xmlsoap.org/wsdl" > Actions taken: December 17, 2002: received issue Issue 5793: CORBA-WSDL/SOAP: Section 1.2.7.4 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: The mapping does not state what value should be used for the default case. Proposed Resolution: Add text stating that any valid value for the discriminant, other than those used in the union definition, can be inserted for the default case. Resolution: Accept proposal from source Revised Text: In section 1.2.7.4 , add the following new paragraph after the existing first paragraph: Actions taken: December 18, 2002: received issues of each otherother Date: Thu, 17 Apr 2003 19:44:43 -0400 From: Tom Rutt Reply-To: tom@coastin.com User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.0.2) Gecko/20021120 Netscape/7.01 X-Accept-Language: en-us, en To: Jishnu Mukerji CC: "corbawsdl-ftf@omg.org" , iwasa Subject: Re: ISSUE 5785 Proposal Based on the discussion below, I summarize the following points: In IDL the TypeID or Pragma Prefix can be used to avoid collisions in module names, as far as repository ids go. The repository id is the core of the CORBA type system. If an IDL artifact (e.g ,) has a #include it must assume that module names are different only if they have a different typeID or Pragma Prefix. Otherwise the must be assumed to be the same module. IDL modules may be defined across several IDL definition artifacts. With our current approach, a WSDL definition document will always use th e namespace. The WSDL port type is bound to a specific url, and that URL is expected to work in accordance with the wsdl definition for its port type. Thus, IT DOES NOT MATTER IF ANOTHER WSDL DEFINITION USES NAMES DIFFERENTLY. Thus, if we start with an IDL artifact, and fully execute all the #includes or imports, by looking at the typeIDs or pragma prefixes we can know if the Module name needs to be qualified by the prefix to assure uniqueness. Thus I propose that we only need to add the prefix for the typeID or pramgaPrefix is we discover the need to do so within a single WSDL Description document. Jishnu Stated: I don't think it is important to include the "IDL:" prefix What is important though is to cover all possible combination of typeprefix and typeid directives. One complication arises because the only way to create a IDL type Rep id with a new version number using these directives is to use the "typeid" directive and in it specify the entire Rep Id including the "IDL:". Also, if there is a typeid directive for a scoped name, then for that scoped name any typeprefix directive that may be active does not apply. This precedence rule is the same as for pragma prefix and pragma id. So here is how we can cover all bases: 1. If there are no pragma or typeid/typeprefix directives that applies to a module name then the scoped name of the module starting from the root module name using the "." character as separator between names shall be the effective Id used in the corresponding WSDL. 2. If there is a pragma prefix or a typeprefix directive that applies to the module name then the prefix specified in the directive/pragma shall be of the form: ":" If no such directives apply to the module name in question then the effective Id shall be the same as in (1). 3. If there is a pragma version that applies to the scopename then the version string shall be appened to the effective Id obtained in (2), that is the effective Id will be of the form: ":" 4. If a pragma id or a typeid directive applies to the module name in question then: 4a. If the first four characters of the string specified in the directive/pragma is "IDL:" then the string with the "IDL:" prefix removed and with the "/" characters in the remaining string substituted by the "." character shall be the effective Id. 4b. If the Repository Id type prefix in the directive/pragma is something other than "IDL:" then ......... 4b choice 1 "the entire id string shall be used as the effective Id" 4b choice 2 "it shall be flagged as an error" 4b choice 3 "the id shall be ignored and the effective Id as computed in (1) shall be used. I think choice 1 (with some munging rule to escape out disallowed characters) is preferable since it allows the use of domain specific Ids. The only problem is that there is the infinitisimal chance of a id clash in certain situations. Comments? Jishnu. Tom Rutt replied,: Tom Rutt wrote: What about the idea of only adding the prefix when a collision is detected in a single WSDL document? Most of the time there will be no collision, thus the names would be quite shorter. I realize that the prefix might be added to a module in some wsdl docs while in others (depending which module does includes from which other module) that same mopule might not have a prefix. However, the schema for each wsdl document will be consistent and without collision. Does it matter that some WSDL documents will use a prefix for a translated IDL module while others will not? Jishnu responded: Whatever is it that we do should be such that the info can be used to be able to create a consistent IDL file in the reverse mapping. I don't know that there is any requirement other than that and that the namespace in the WSDL document should be unique. Jishnu. S -- ---------------------------------------------------- Tom Rutt email: tom@coastin.com; trutt@fsw.fujitsu.com Tel: +1 732 801 5744 Fax: +1 732 774 5133 Date: Tue, 29 Apr 2003 13:59:02 -0400 From: Tom Rutt Reply-To: tom@coastin.com User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.3) Gecko/20030312 X-Accept-Language: en-us, en To: corbaWsdl ftf , tom@coastin.com Subject: Final Proposal Wordsmith for Issue 5785 This is a new proposal for resolving Issue 5785, the module clash issue. If there are no comments, I will send this out with Vote 2 on Thursday (time is running Short, the three week rule is due May 12). This only applies the prefix if a collision occurs. Please review this and give comments ASAP. -- ---------------------------------------------------- Tom Rutt email: tom@coastin.com; trutt@fsw.fujitsu.com Tel: +1 732 801 5744 Fax: +1 732 774 5133 xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:st1="urn:schemas-microsoft-com:office:smarttags" xmlns="http://www.w3.org/TR/REC-html40"> Issue 5785: CORBA-WSDL/SOAP: Section 1.2.2.1 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com) Nature: Uncategorized Issue Severity: Summary: Section 1.2.2.1 - tns is a single namespace for all mapped IDL from the entire corba world. The module names for IDL across this vast namespace, need to be sufficiently qualified with pre-fixes (such as Type ID, or Pragma Prefix) to avoid module name collisions. Proposed Resolution: FTF needs to determine appropriate prefixes or namespaces to assure module name uniqueness in the generated WSDL Discussion: The brute force approach is to use the scheme that is used for IDL form of repository ID for a Module. Prefix the module name with the string from the pragma prefix IDL directive, usin g (i.e. An OMG IDL pragma of the format: #pragma prefi x ) After discussion, it was realized that the collision is only important to resolve within a single WSDL document. Also if no collision occurs, then the extra prefix is not required. Proposed Resolution: Append a prefix (if present and required) in front of the module name. Need to fix all examples to use this pragma prefix in module name prefixes Proposed Revised Text: Remove the following text from th e row of the table in 1.2.2.1 (There are still some issues to be addressed at the FTF stage around how we should map repository IDs into namespaces, so it is likely that some sort of concatenation of namespace and ID will occur). Also, in section 1.2.4 Modules: Change IDL modules are represented in WSDL as prefixes to the names contained in the module. These prefixes are concatenated using a as a separator. For example, an entit y in a modul e will have the nam e in WSDL. to the following: 1. For every module name which does not collide with another module name, across the set of IDL artifacts which are being mapped to a WSDL document, then the scoped name of the module starting from the root module name using the "." character as separator between names shall be the effective Id used in the corresponding WSDL. 2. If there is a module name collision, apply the following rules to the module name which causes the collision: If there are no pragma or typeid/typeprefix directives that applies to a module name then the scoped name of the module starting from the root module name using the "." character as separator between names shall be the effective Id used in the corresponding WSDL. If there is a pragma prefix or a typeprefix directive that applies to the module name then the prefix specified in the directive/pragma shall be of the form: ":" If no such directives apply to the module name in question then the effective Id shall be the same as in (1). If there is a pragma version that applies to the scopename then the version string shall be appened to the effective Id obtained in (2), that is the effective Id will be of the form: ":" If a pragma id or a typeid directive applies to the module name in question then: i. If the first four characters of the string specified in the directive/pragma is "IDL:" then the string with the "IDL:" prefix removed and with the "/" characters in the remaining string substituted by the "." character shall be the effective Id. ii. If the Repository Id type prefix in the directive/pragma is something other than "IDL:" then the entire id string shall be used as the effective Id Fix all examples < since there are no module collisions in the example, this is unnecessary> Actions taken: December 17, 2002: received issue Date: Tue, 29 Apr 2003 15:28:51 -0400 From: Jishnu Mukerji Organization: Software Global Business Unit, Hewlett-Packard User-Agent: Mozilla/5.0 (Windows; U; Win98; en-US; rv:1.0.2) Gecko/20021120 Netscape/7.01 X-Accept-Language: en-us, en To: tom@coastin.com Cc: corbaWsdl ftf Subject: Re: Final Proposal Wordsmith for Issue 5785 Tom, Overall, looks good. Do we need to say something about mapping characters that might appear in the typeid/typeprefix/pragma strings that are not legal characters in the name of a namespace? I don't know for sure that there are such possibilities, but we should either make sure that there aren't, or provide a character munging specification for those characters that are problematic. Thanks, Jishnu. Tom Rutt wrote: This is a new proposal for resolving Issue 5785, the module clash issue. If there are no comments, I will send this out with Vote 2 on Thursday (time is running Short, the three week rule is due May 12). This only applies the prefix if a collision occurs. Please review this and give comments ASAP. ------------------------------------------------------------------------ Issue 5785: CORBA-WSDL/SOAP: Section 1.2.2.1 (corbawsdl-ftf) Click here for this issue's archive. Source: Fujitsu (Mr. Tom Rutt, tom@coastin.com ) Nature: Uncategorized Issue Severity: Summary: Section 1.2.2.1 - tns is a single namespace for all mapped IDL from the entire corba world. The module names for IDL across this vast namespace, need to be sufficiently qualified with pre-fixes (such as Type ID, or Pragma Prefix) to avoid module name collisions. Proposed Resolution: FTF needs to determine appropriate prefixes or namespaces to assure module name uniqueness in the generated WSDL Discussion: The brute force approach is to use the scheme that is used for IDL form of repository ID for a Module. Prefix the module name with the string from the pragma prefix IDL directive, usin g (i.e. An OMG IDL pragma of the format: #pragma prefi x ) After discussion, it was realized that the collision is only important to resolve within a single WSDL document. Also if no collision occurs, then the extra prefix is not required. Proposed Resolution: Append a prefix (if present and required) in front of the module name. Need to fix all examples to use this pragma prefix in module name prefixes Proposed Revised Text: Remove the following text from th e row of the table in 1.2.2.1 Also, in section 1.2.4 Modules: Change. as a separator. For example, an entit y in a modul e will have the nam e in WSDL. to the following: 2. If there is a module name collision, apply the following rules to the module name which causes the collision: 1. If there are no pragma or typeid/typeprefix directives that applies to a module name then the scoped name of the module starting from the root module name using the "." character as separator between names shall be the effective Id used in the corresponding WSDL. 2. If there is a pragma prefix or a typeprefix directive that applies to the module name then the prefix specified in the directive/pragma shall be of the form: ":" If no such directives apply to the module name in question then the effective Id shall be the same as in (1). 3. If there is a pragma version that applies to the scopename then the version string shall be appened to the effective Id obtained in (2), that is the effective Id will be of the form: ":" 4. If a pragma id or a typeid directive applies to the module name in question then: i. If the first four characters of the string specified in the directive/pragma is "IDL:" then the string with the "IDL:" prefix removed and with the "/" characters in the remaining string substituted by the "." character shall be the effective Id. ii. If the Repository Id type prefix in the directive/pragma is something other than "IDL:" then the entire id string shall be used as the effective Id Fix all examples < since there are no module collisions in the example, this is unnecessary> Actions taken: December 17, 2002: received issue -- Jishnu Mukerji Senior Systems Architect 1001 Frontier Road, Suite 300 Technology Office Bridgewater NJ 08807, USA Software Global Business Unit Tel: +1 908 243 8924 Hewlett-Packard Company Fax: +1 908 243 8850 mailto: jishnu@hp.com Date: Tue, 29 Apr 2003 17:19:53 -0400 From: Tom Rutt Reply-To: tom@coastin.com User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.3) Gecko/20030312 X-Accept-Language: en-us, en To: Jishnu Mukerji CC: corbaWsdl ftf Subject: Re: Final Proposal Wordsmith for Issue 5785 Jishnu Mukerji wrote: Tom, Overall, looks good. Do we need to say something about mapping characters that might appear in the typeid/typeprefix/pragma strings that are not legal characters in the name of a namespace? I don't know for sure that there are such possibilities, but we should either make sure that there aren't, or provide a character munging specification for those characters that are problematic. According to Essential XML quick Reference (I could not find the normative text in the XML schema docs): " Element names in XML are case sensitive and must begin with a letter or an underscore (_). The initial character may be followed by any number of letters, digits, periods (.), hyphens (-), underscores, or colons (:). However, because colons are used as part of the syntax for namspaces in XML, they should not be used except as described by that specification. Element names that begein with the character sequence "xml" or any recapitialization therof, are reserved by the XML specification fo future use. " Thus we would need to change anthing not in the list above (in the name of the element as mapped from IDL) to one of the above. Perhaps substitute bad chars with underscore. What can occur in typeID or pragma prefix (or module name or and valid IDL name) which is not above? Tom Rutt Thanks, Jishnu. -- ---------------------------------------------------- Tom Rutt email: tom@coastin.com; trutt@fsw.fujitsu.com Tel: +1 732 801 5744 Fax: +1 732 774 5133