[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Minutes of Atlantic UBL TC call 19 October 2005
MINUTES OF ATLANTIC UBL TC MEETING 15H00 - 17H00 UTC WEDNESDAY 19 OCTOBER 2005 ATTENDANCE Peter Borresen Jon Bosak (chair) Marty Burns Tony Coates Mavis Cournane Michael Dill Mike Grimley Betty Harvey Anne Hendry Sue Probert Zarella Rendon Andy Schoka Paul Thorpe Sylvia Webb The meeting was quorate. STANDING ITEMS Additions to the calendar: http://ibiblio.org/bosak/ubl/calendar.htm None. Liaison report: ebBP TC JB: Note that we have been asked for input. Subcommittee report: TSC AS: A dial-in capability will be provided at this week's TSC f2f. Team report: Code Lists TC: I have an example that SG sent me today. I have 1.0 schemas, and will have (1) original style, (2) 2.0 style enumuerated/validated, and (3) 2.0 style unvalidated; will see what happens with the different example. Not sure exactly what GKH is doing, but don't think we're working at odds. Will engage GKH after am done with latest. Review of Pacific and Europe/Asia calls No comments. Schedule review PSC: SW: We have the content, it's just not clean enough yet; SG hopes to be able to work on this over the next week, but it's highly unlikely we will make the 24th. Catalogue: JB: I haven't been given a reason to think that catalogue will slip. index.htm: JB: How necessary is it to talk about methodology the way we did in 1.0? SW: Just point to previous writings. AS: It's useful to know... so include a link to 1.0. ACTION: JB to create the first draft of index.htm. ACTION ITEM REVIEW ACTION: TC members visiting NYC in October or November to check out the Sun facilities and see whether they are suitable for a UBL TC meeting. JB will be coming through in mid-November and will also check then. Pending. ACTION: JB to create a schedule for UBL 2.0 supplementary materials. Pending. FOR THIS MEETING UBL vice chair AGREED: Tim McGrath is appointed the UBL vice chair to replace Mark Crawford, who has had to leave active membership due to a change in employers. NDR editors AGREED: Mike Grimley is appointed Lead Editor of the NDR document. Mike and Mavis will both be listed as editors when the document is published. UBL QDT module From Pacific TC minutes: Problem: An unintended consequence of adopting the ATG UDT is that the data model is not compatible with the UBL UDT model, so we will need a UBL QDT that matches the ATG UDT. This would have been an SSC task, but TonyC, SG, AH are all having problems attending. SW: Needs to be done urgently; this is in the critical path for schema generation. The problem is that the ATG UDT starts at the ACC level, while UBL starts at the BIE level. As a result, the model structure is not the same. QDTs are based on UDTs; we do not have a UBL UDT model, and in previous discussions in the PSC, TM and others were opposed to creating one, wanting to use the ATG model, which is not in spreadsheet format but exists only in Edifix. So we cannot create UDTs. Since the UDT struct is coming from CEFACT unchanged, we now need a QDT model that is the same in structure. MC: So would we need new NDRs for UDT schema? SW: We did not see any changes to the rules, just the number of spreadsheet columns and where data is put in those columns. There is no spreadsheet version of the ATG model; our users will need one. JB: We're not providing spreadsheets of the UDT, so why do we need one for the QDT? SW: If not provided as a spreadsheet, we would need to know that it won't change. JB: So we either create one hardwired version through EF programming or we have to come up with a spreadsheet format for this. SW: Yes, and note that the QDT is something that gets added to. JB: So we need a spreadsheet format for this. Volunteers? BH: Not a CCTS expert, but willing to help. Does this require a tech person? SW: Yes; you need to have some understanding of CCTS 2.01. The ATG2 version is a CCTS data model; looks like UML class diagrams. Our fallback is to do this programmatically, but this would prevent further changes and would take longer than a week. ACTION: JB to appeal to the list for volunteer(s) to develop a QDT spreadsheet. Code lists From Pacific TC minutes: Question: are we going with the UBL 1.0 code list structure for the code lists that were in UBL 1.0? Or do we use the structure of the code lists that we have agreed to adopt from ATG2? JB: What's the difference? ACTION: SW to investigate the difference between the structure of the enumerated code list schemas we've agreed to incorporate from ATG and the ones that were included in UBL 1.0 and provide her findings by the end of week. Minor versioning: see MG's issue list: http://lists.oasis-open.org/archives/ubl/200510/msg00046.html JB: Note answers to MG's questions 3 and 3a from EduardoG: http://lists.oasis-open.org/archives/ubl/200510/msg00050.html AGREED to accept EG's answers to 3 and 3a. MB: We will need substitution groups in order to implement the original customization document. JB: Noted; relates to MG's question 4. MC: We also have a straw proposal from Bill Burcham [off the list]. It assumes that we're doing versioning from the model rather than the schema and is based on spreadsheets. JB: So this is an additional question: Shall the models be normative? MC: Another question is: Do we still want to do minor versioning? JB: Well, we have to do minor revisions; the question is what to call them. If we can't figure out how to do minor versioning properly speaking, then we'll either assign every version an integer or we'll change what we mean by fractional version numbers. Discussion of MG's Question 2: Should the namespace change for minor versions? MG: Agree with Mark Crawford: as long as it's backwards compatible, the namespace URN should not change; this makes it easier for implementers. MB: Agree; if the difference is just extensions, you should not be penalized if you've already implemented the major version. JB: There are two sides to this. If the function of a namespace is to mark semantic distinctions, which was the original idea, then clearly you don't change the namespace for a minor version, because everything that was in the major version still has the same meaning. This is how I was leaning before a recent discussion with Nikola Stojanovic of RosettaNet. RN discussed this in depth and came down strongly on the side of assigning each minor version its own namespace. The reasoning is that a namespace URN tells a processing application what to expect. If the URN says version 2.0, the app expects not to be surprised by elements that are not in 2.0. So a 2.1 document that says "2.0" is basically lying to the app about what it can expect to see. MG: The assumption is that the app should ignore an element it doesn't know about. This is just good practice. MB: Like other apps. TC: But the fact is that namespaces are overwhelmingly used to provide version information about schemas. There's no other standard way to do this. MB: Schema location is used for this. TC: No, the location is generally thrown away. JB: And replaced by the location on the machine where the schemas are installed, for example. TC: So really, there's no other way to tell. BH: How about a version attribute? TC: But that's not standard. In practice, there is no standard piece of the infrastructure that looks at a top level attribute and knows to use that schema. We would be asking everyone to use the structure we define. JB: We could have an attribute that at least tells UBL applications what to expect. ??: It would help to know what other widely used schemas do about this. We know that RN changes the URN to reflect minor versions and we've heard that OAGIS does not. ACTION: JB to post a question on ubl-dev: What do other widely used schemas do about this? ACTION: NDR editors to post a revised issues list including the new question about which is normative (model or schema) and information about BillB's proposal. The next time that both NDR editors can attend the call is 2 November; we will schedule a full discussion of minor versioning for that meeting. MD: Concerned that someone in another environment can carry the same semantic information, for example using RELAX NG. JB: We already do that with ASN.1 and would like to see more, but there can only be one normative form of the schemas (relates to question to be discussed 2 November). Jon Bosak Chair, OASIS UBL TC
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]