[tmql-wg] TMQL Problem (scoping occurrences).

[Michael Chapman]

As before this is more a TM problem than a TMQL problem, but if I can't get a 
good TM I won't be able to query it, so ... ...

Also this may (but I doubt) be unique to LTM, but as LTM is so intertwined 
with TMQL now, then that seems to make it a general problem (if problem it 
really is).

O.K.

If I have:

{rodin, description, [[Sculpteur et aquarelliste 
		(Paris, 1840 - Meudon, 1917)]] } /lang-fr
{rodin, description, [[Sculptor and watercolour painter 
		(Paris, 1840 - Meudon, 1917)]] } /lang-en
{rodin, description, [[Auguste Rodin (1840-1917) was arguably
		the leading sculptor of the late Nineteenth
		and early Twentieth Centuries. His most 
		famous works include ... ... ]] } /lang-en

and I want to present my data, then (to my mind) a good rendering would be:
a) everything;
b) just French, or just English; or,
c) everything _excluding_ mere language duplicates
	e.g. if I ask for 'full', but English preferred, I get the second and third
	if I ask for 'full', but French preferred, I get the first and third.

Without making the occ-type unique (in practice difficult!!,) as in 
description-001 (used for the first _and_ second above), description-002, 
etc. there seems to be no easy solution.

(One can argue a similar line about URLs were two URLs are language variants 
of identical material, whilst a third is the same text but with illustrations 
(or illustrations in colour rather than black and white, or ... ) ... that is 
one wants to hide multiple examples that only differ in language, but keep 
any with other differences.

One solution would seem to be:

{rodin, description, [[Sculpteur et aquarelliste 
		(Paris, 1840 - Meudon, 1917)]] /lang-fr ;
		[[Sculptor and watercolour painter 
		(Paris, 1840 - Meudon, 1917)]] /lang-en
		}
{rodin, description, [[Auguste Rodin (1840-1917) was arguably
		the leading sculptor of the late Nineteenth
		and early Twentieth Centuries. His most 
		famous works include ... ... ]] /lang-en
		}

which is surely right, afterall it is the resource that is (in this case, at 
least) scoped, not the whole occurrence (?).
The above syntax also reflects that for topics:
[ lond = "London" /lang-en ; "Londres" /lang-fr ]

I can't be the only person to have hit this problem, can I ? If so is it real 
and what do others do?

Regards,

	Michael Chapman

PS Obviously real examples with several languages in the database (but not 
consistently), complex language preferences of users and all, get much more 
complicated, but I hope the trite example is enough to make the point.
As with all trite examples one can see easy 'work arounds' ... but that is 
another matter. 
The problem is also -obviously- more general than just  languages.  MC

-- 

interarb
   Culoz, France
   tel: 00+33-4.79.87.16.30
   fax: 00+33-4.79.87.16.33
interarb is an information provider in the subject areas of arbitration 
and dispute resolution, it is dedicated to the supply of essential 
support to practitioners and users of international dispute resolution 
services on a global basis. 
   <http://interarb.com>
interarb is a trademark