Re: [AUCTeX-devel] Re: Interface of kpathsea file search

[David Kastrup]

Ralf Angeli <address@hidden> writes:

> * David Kastrup (2005-11-23) writes:
>
>> Ralf Angeli <address@hidden> writes:
>>
>>> As far as I understand the variable's doc string and intention the
>>> first item in each element is rather something like a format as used
>>> by the kpse* tools, e.g. as in `kpsewhich -progname latex -format tex
>>> article.cls'.
>>
>> It is exactly "something like": we use it as a clue to detect what
>> kind of search would be wanted from kpathsea.  Maybe we should indeed
>> use progname instead of TEXINPUTS.latex, but I don't know what the
>> MikTeX substitutes will actually do in either case.
>>
>>> But we are using it as file extension.
>>
>> We are _mapping_ it to a file type and a set of extensions.
>>
>> I am not saying this is pretty.  It is merely trying to use
>> kpathsea transparently without changing all the callers of the file
>> search functions.
>
> But the callers pass file extensions to the search function(s).

> Usually `TeX-search-files-kpathsea' is called through
> `TeX-search-files' and `TeX-search-files' accepts only a list of
> extensions, not some abstraction of it.

I am afraid I don't get your point.  What "abstraction" are you
talking about?

> So the _intention_ of `TeX-kpathsea-format-alist' is a mapping of
> something like a format specifier

There is no such thing "like a format specifier".
`TeX-kpathsea-format-alist' just takes the first element of the
extension list as a heuristic what to look for.

> to a file type and a set of extensions but the actual usage differs.

The one thing that seems somewhat odd is that
`TeX-kpathsea-format-alist' also provides an extension list again.  I
am not sure that the original extension list would not be sufficient
for that purpose.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum