UBL LCSC Status Meeting Minutes

Thursday - 16 January 2003

Attendees: Bill Meadows, Anne Hendry, Sue Probert, Tim McGrath, Lisa Seaburg, Marion Royal, Jon Bosak, Sally Chan.

Proceedings:

- Yesterday Gunther sent out global element referrencing While they looked fine at beginning, we didn't realize implication.

- Where we used to be there was party (complex type) id would be element wihtin it. All metadata, proterpties, things in spreadhsheet columns would be right there where element was defined. Under revised schema, with a complex type called party, there is id, but it is only a reference - the id is actually defined elsewhere globally. This is problematic. Now need to call it partyid, can't call it just id. Gunther's argument is that we'll end up with huge tag names - reduces point of having global elements.

Sometimes we have a tag name that is reused - cityname is reused because address is reused. In that case cityname could be a reference to something somewhere else. However the tag name called id is not uniques we do the same with codename, etc - lots of them - the primitive types.

Option is that we try to make the names unique (becomes the dictionary entry name?) or try to figure out a different way to do it. Whole purpose of having tag names was to qualify. Trouble with this now`is that it doesn't show the context?

Tim sent out two sample schemas yesterday. 1 is quick and nasty solution 1025_tim_draft. Took Gunter's original, removed annotation from bbi elements (if party referenced id, and looked at def of id, def of id said 'this is a core compoennt identifier, w/o contect or further definition). By removing def we got away from the quesiton of what def do we use. This removed innacuracy of orignal schemas. The way it was working was the last time it was referenced was the definition it had. Definition is still within the schema, but the definition is where the party id is referenced, and immediately following it is the def of id as it is used in that context (w/in party). But if you follow that reference, there isn't anything there becasue we don't know which one to use. Gunter didn't like that. Anyone have other ideas?

- Library 07 reusable is slightly more sofisticated option. Gunter's criticism is that if you were user of UBL and wanted to construct your own aggregates, you couldn't look at the BBIEs and know what you are looking at. Inside there are generic things like name, id, code. Then there are also name, cityname, account code, whnich are unique, but used several times (in reusable components). We attempted to (in second schema) produce objects that weren't synonyms. Property and object class were the same. works for cityname but not for transportid, or others ... So this is a cluged attempt to resolve this, until we agree how to resolve. This is making it hard for us to complete the xsds.

Lisa - Big diff between this and what Gunter sent out in last few` hours. This is 04. lp7 is alot more reusable. Are all rthese versions coming off same spreadhsheet? No, there were a few fixes (occurrance columns, etc), but no def changes or adding/subtracting of actual elements. This now is just different ways of looking at them. Mike needs to know this - it's all the same contnet, just diff ways of looking at it.

Sue - For first time we have a decent set of contents for what we discussed in 2001. It was theoretical then, now we have real data. Then we were discussing if business content was correct. We're just going through stages of development and we're going very fast and accomplishing a lot. We just need to get the release out. Suggest that we keep the frozen spreadhsheet file.

Tim - we fixed errors, and there has been a change log. We're not exprimenting. The spreadsheet has typographical errors that have been fixed - that's all. The logical content (structures, definitions, etc) are still the same. Only errors (wrong thing being copied into cells, etc). To get release out, next week even, we need to address this today.

Gunter's solution is more elegant, but you end up with very long names. He's followed the NDR rules now. But is this usable? From point of view of having a technical review of material, we don't have anything so ugly as to be unusable. Only down side is we have long names. Don't forget we have Jon, Mark, Eve showing them how this is used. Stands behind what we've said.

Jon - not surprised that we have something horribly verbose. But difference of opinion. Best thing is to go with what NDR voted for and let world decide. There are points where you need to let the world decide. Eve was slightly surprised that we'd go ahead and make this change at this time. Need to decide if this is doable technically. Can go out with cleaned-up version, or can go out with NDR plan of record.

Tim. Problem is that NDR is still undecided. There is reusable_4a, reusable_global - two diff ways and are still under debate. In 4a, referrencing is by type. No-one is sure which is best. This is what would be good to debate at plenary.

We should pick one for getting the release out. Today we can go out supporting the NDR rules. Which one do we do with. Which is closest to the one NDR has asked Gunter to do. Lisa is NDR at this meeting. It looks like 04_global, but haven't looked at 4a enough to understand. Tim - agrees. In spirit of the NDR paper, o4_global is the one, but this means we've thrown out the ... rules. Lisa - global/local is much more importand thatn truncated, etc rules.

Will we have time to have samples? Yes, we will have samples, but maybe not by Monday.

In order for samples to work need to extend element names in spreadsheet. But can't find 1:1 mapping. Can use sample building as excercies to illustrate issues. Don't ahve to solve them initially. Need to build complex extension methodology.

Schemas: what will it take to take the 04_glboal and generate schemas off that. This could be done in a couple of hours. This is good, then samples could be built from that. But nervous because this is where we sat yesterday and then after drilling down ran into problems. On first glance this looks usable. Looks pretty good using spy (can use schema mode) but don't see 1/2 of definitions. But beyond having ugly tag names, o4_global looks fine - is closer to the NDR rules. Put priority on global/local, rather than naming. Bill is concerned because doesn't follow naming truncation rules. Goes without truncation. Eve's suggestion has been incorporated - no text (phone.text just becomes phone) Things have to be qualified by their object class - using shortened dictionary name. Haven't abandonded shortening rules, just not using them completely.

Jon would like a set of slides describing the current implementation and issues for his presentation.

What will we do for ATG? Can we introduce this then? Mark's been working on having version of the NDR paper for next week. If we could get some of the sample schemas from point of view of showing the issues. Gunter has two arguments with 4a and 4global. Issue is when you try to build other docs on this. This is what Tim will do next. He'll let us know by tomorrow how this works.

Tim will send validateion team the results of today's work. Will get sent to list and to validation team. This is great that we're at this stage!

Advisory for QA team (call tomorrow morning, early). Need wrapper paper. Scope doc - is it in that doc now? Yes, because ISO template uses 'scope' to discuss scope of docuement section, this is renamed. Section 7 contains xsds. Section 7.3 is the first copy of paper on UR language (extension methodology). NDR has been slowing rewriting to change name UR. Lisa will send to Tim tomorrow morning. This was called 'release notes'.

Marion - is there any way we can get the text that Jon just used to describe the issues. Tim, can we create a section called 'release notes'? By putting this in 7.3 it may not jump out enough at people. Would this be separate document? Dont' have to adhere to template. At this point could be a separate section in front of introduction. Good - agreed. Like idea of having one umbrella doc that refereences the whole thing.

Lisa - should we say something about going with cleanest global verions we can do and doesn't adhere to all naming rules? Tim could compile excerpts from emails to describe the current release issues. Will frame positively.

Marion - closure on QA. Will focus on doc Tim sends out jan16th library content review.

Next meeting will be tomorrow at same time. Then next week will be Monday meeting (at Tuesday time).