>>Date: Thu, 14 Feb 2002 10:08:11 +0100 (CET)
>>From: Guenter Milde <[EMAIL PROTECTED]>
>>Subject: Re: Re: customizing natbib
>>To: "Jean-Pierre.Chretien" <[EMAIL PROTECTED]>
>>
>>On Wed, 13 Feb 2002 18:08:07 +0100 (MET) wrote "Jean-Pierre.Chretien"
<[EMAIL PROTECTED]>:
>>
>>> >>> I just checked out this website. It seems like the amsrefs
>>> >>> package is meant for publishing mathmatical documents, something
>>> >>> I won't be doing. Also, if I understand it correctly, amsrefs
>>> >>> doesn't wouldn't offer any flexibility. The database of your data
>>> >>> base has to be what you would use in a latex document. What
>>> >>> happens if you need to change one element in this database?
>>> >>> Wouldn't you have to every single entry by hand?
>>
>>The discussion ispired me to take a closer look at the amsrefs page...
>>
>>Actually, amsrefs is even more (and for a puer latexer also easier)
>>configurable (i.e. more flexible) than the traditional bibtex.
>>OTOH, it is really new, so there are still not many ready-made styles
>>available.
>>
>>> >>2) Let me see from the example document (jktest.ltb):
>>> >>
>>> >> \bib{MR58:27738}{book}{
>>> >> author={Andrews, G.~E.},
>>> >> title={The {T}heory of {P}artitions},
>>> >> publisher={Encyclopedia of Mathematics and its
>>> >> Applications, Vol. 2,
>>> Addison-Wesley,
>>> >> Mass.-London-Amsterdam},
>>> >> date={1976},
>>> >> }
>>>
>>> This is a very bad data structure from the point of view of
>>> data manipulation and indexing because the fields are not atoms:
>>> the publisher field mixes the collection name, the volume number,
>>> the publisher name and the publisher address.
>>
>>Actually, it is just a bad example! Besides the mixing of different fields,
>>it still shows the "escaped capitalization "{T}heory" which is not necessary
>>to have with armsrefs
>>(As it works the opposed way to bibtex:
>> Bibtex: give title All-Capitalized -> some styles downcase it
>> AMSRef: give title First-Capitalized -> some styles capitalize it
>>
>>Here is a better example (from AMS-refs/doc/latex/amscls/niklas.ltb), that
>>shows that there are enough fields to have an atomic structure (BTW also new
>>fields could be defined by a new style and defining a new style is basically
>>writing a new LaTeX package):
>>
>>\bib{fn95}{article}{
>> author={Farnsworth, K. D.},
>> author={Niklas, K. J.},
>> date={1995},
>> title={Theories of optimization, form and function in branching
>> architectures in plants},
>> journal={Functional Ecology},
>> volume={9},
>> pages={355\ndash 363},
>>}
>>
>>> It is very easy to build up such a command from a .bib file, but the reverse
>>> is not possible. Let's stick to .bib syntax
>>> as specified by O. Patashnik and enforced by a lot of various
>>> data manipulation and indexing tools (I personnaly use Nelson Beebe's
suite).
>>> Then write an alternative to bibtex which transcodes .bib standard
>>> to amsref standard.
>>
>>Actually, such a program is already in the amsrefs suite, i.e. you can use
>>your existing Databases with amsrefs :-)
In fact, I wanted to stress the fact that database typing
inside documents (which is a current practice) should be avoided
because a bibliography tag is by definition an piece of data
which should be written once correctly forever, and available
autonomously for copy (that's why I don't like some facilities like the
STRINGS or crossref features: the bib record looses autonomy
ant can't be simply copied for reuse).
So the question is only about the choice of the standard:
- using latex in an abstract field (which does not exists in the original
bib data structure (which knows only about note field AFAIR) or in titles
(which I use often) imports (La)TeX inside this piece of data
this is clearly not recommended for portability even if the output
is much better; this must be the reason why .bib is not a standard for
citation outside the academic world using LaTeX/LyX;
- using on of these standards (among which a lot are proprietary AFAIK)
needs conversion to bib or amsref, if ever possible without
licensing.
My feeling is that the most complete format must be retained, and
its seems easy to regain portability by degrading the latex constructs
by postfiltering. For this particular point, I would say that the .bib
file is a de facto standard from which others should be built,
and I think that for once, it's up to "others".
But if amsref is OK from this point of view, it's great,
we agree: just use the amsref builder instaed of bibtex :-)
--
Jean-Pierre
