RE: a new UID discussion

[Harald Hauschildt <hha@mosaic-ag.com>]

Hi Scott,

the idea is not to store 10000 similar tag names, the idea is to 
store just ONE meaning
of e.g. "family name" and address this meaning to a UID 4711.

A SO than browseres trough the (local) repository and checks for the 
meaning of their own tag <lastname>
and expand the message description (DTD or schema) with <lastname UID 
4711> (or only <UID 4711> if they want).

Any ebXML user is able to do their very one mapping  by comparing 
their tags meanings against the meanings of the UIDs stored
at the repository (here the human readable interface is of course 
very importnant, but your are doing this process only once).

Again: The mapping of UID 4711 to a tag like <name3> or <lastname> is 
done by a ebXML user OUTSIDE the repository!!

Harald

-----Original Message-----
From:	Nieman, Scott [SMTP:Scott.Nieman@NorstanConsulting.com]
Sent:	Wednesday, November 15, 2000 2:17 PM
To:	'Harald Hauschildt '; ''ebxml-architecture@lists.ebxml.org' '
Subject:	RE: a new UID diskussion

>> (e.g. map <lastname UID=4711> to <name3 UID=4711) regardless of my 
tag-name within the message)

This is nice, but introduces unnecessary registry overhead which core
components is suppose to solve.

Also, the SO needs to tell the world that name3 means 4711; the 
registry
can't do that auto-magically.  Or are you suggesting that in the
registration process, the registry looks at name3, parses "name" out 
of it,
determines 10,000 similar names and the SO is suppose select the 
equivalent.
I hope not!  Tell us how this is supposed to work if the SO does not 
do it.

>>My proposal (to the TA-Group at this point) is to claim the above 
UID
handling as a indispensable necessity...!

Which proposal is this???  Parsing and GUID'ing everything?  The 
compromise
is that the SO specifies the level of granularity.

Scott