OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

ubl message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]


Subject: Re: [ubl] UBL master data for maintenance


Hi Micheal,

We are following our original agreement, which included maintaining the
ss which could/would be used for modeling.

In order to make this happen without the problems we saw earlier, the tc
agreed to do an extensive alignment excercise between the curren 1.0 ss
and the schemas produced by ef for 1.0.  Through discussions between David
and Stephen, and from David's input at the SF F2F, the SSC has been working
on this alignment.  Many of these issues have been pushed up to the TC and
have been discussed at each meeting.   Once we have those issues resolved
there should not be any problem with ef taking in ss for new changes
(which was was agreed, at the f2f in copenhagen), and for ef to generate
new ss (which would be a bi-product of this process).

We are almost done with thie alignment process.

-A

>Dear Jon,
>  As far as I remember, we have at least twice picked up about 200 technical
>errors from these UBL spreadsheets. And everybody knows that the formulas of
>the spreadsheets in order to build a Dictionary Entry Name did not work
>correctly. We even have the case that the spreadsheets have a) redundant
>data and b) data on the wrong place, just in order to meet the requirements
>of formulas, whereas it should be just the other way round. We now know that
>handcrafting is simply not capable of producing quality results for huge
>amounts of data.
>
>  Therefore, if we know that a given procedure produces errors, then the
>procedure has to be changed.
>
>  After reading the UBL call minutes, I understand that the current decision
>is to view spreadsheets as the maintenance master data of UBL. This means
>necessarily that UBL plans to produce further errors in the future too. I
>would like to ask for which reasons the earlier decision to use Edifix as
>the master has been overturned?
>
>  I think, our (GEFEGs) history of contribution A) in developing
>functionality to produce more and more complex UBL schemas, B) in supporting
>the NDR development by running an implementation verification and C) in
>delivering a QA instrument by reading/producing SS is quite a lot for the
>volunteer work of a small team.
>
>  As I strongly believe that nobody wants me to do unnecessary work by
>correcting errors, which are in fact avoidable, I'd like to put on the
>agenda for one of the next calls, the question of how UBL intends to deal
>with the necessary QA to remove the handcrafted errors each time the
>'master' . SSs need to be reimported into the edifix tool (EF). In
>particular, who is going to do this work which consists of importing into
>EF, going through laboriously correcting all the reported errors and
>exporting again and which may require, on previous experience, more than one
>iteration.
>
>  BTW: if UBL was to be really serious about production quality, we should
>need to check through two tools in order to have enough possibilities of
>double checking the resultant schemas. This has always been the best
>implementation verification.
>
>  Once Tim said to me, that he was astonished that he has not yet seen me
>become as frustrated as Gunther and then Chi-Kai were. However, it seems
>that frustration is again being introduced - is this intentional?
>
>  Best regards,
>  Michael  Dill
>  GEFEG mbH




[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]