Re: Bibulus DTD: An concrete example

Torsten Bronger <[email protected]> writes:

> Thomas Widmann <[email protected]> writes:
>
>> Torsten Bronger <[email protected]> writes:
>>
>>> [...]
>>>
>>> Is there any way to avoid creating a list of city names, at least
>>> for the most important, say, 50?  I don't see any.
>>
>> No.  But I think it's important to note that the list probably will
>> be different from language to language.
>
> Okay.  And where and how are the lists hosted?

For each language, there is a file in the Bibulus directory (e.g.,
English is stored in Bibulus/en.pm).  I've only put in a few city
names so far, exactly because the required format wasn't obvious.

> I withdraw my statement: They should be XML in order to allow for
> automatic generation of the lists for the documentation, and maybe
> for the semit-automatic generation of XSLT/Perl/Whatever code for
> applying them, too.

That makes sense.

>> [...]
>>
>>> *or* they have to write "Moscow".
>>
>> Personally, I dislike falling back to English.  It's bad enough
>> that tags have to be in a particular language.
>
> I dislike that too, but the heuristics involved in finding the
> proper language are dangerous.  If the whole entry, or the whole
> bibliography has an xml:lang=3D"ru" the author can (must) write Moscow
> in Cyrillic.

Agreed.

>> [...]
>>
>> Plenty of cities wouldn't have American English variants.  Let me
>> propose another way to deal with it:
>>
>> [...]
>>
>> Another example (with fewer explanations):
>>
>> XML: <city country=3D"ie" xml:lang=3D"da">Dublin</city>
>> Internal: -> Baile =C1tha Cliath/ie/ga
>> German context: -> Dublin
>>
>> Does that make sense?
>
> Yes, but it is more about the implementation rather than the
> standard itself (the file format).

Yes.  I think that if we know what the implementation requires, we can
come up with a good file format fairly quickly.

> The question is how to deal with tags when no xml:lang is set at
> all.  I'd suggest American English, since this seems to be an
> unwritten law in XML.  If there is no English name for the city,
> well, then you have to use xml:lang.

Wouldn't it be better to produce an error message if no xml:lang
attribute is set anywhere?

> [...]
> Things like the "citation index" can only work with standardised
> names, so I think we can take them for granted for our
> considerations here.  The only critical thing is whitespace in the
> abbreviation.

Also whether periods are used or not (Phys Jour vs Phys. Jour. vs
Phys.Jour.), as far as I know.

Another related topic: Do we also need to maintain a list of
institutions (e.g. Aarhus Universitet =3D University of =C5rhus)?

/Thomas
--=20
Thomas Widmann          Bye-bye to BibTeX: join the Bibulus project now!
[email protected]                                <http://www.bibulus.org>
Glasgow, Scotland, EU     <http://savannah.nongnu.org/projects/bibulus/>