[tex-live] bibtex8

Gerben Wierda Gerben.Wierda at rna.nl
Thu Oct 9 11:44:28 CEST 2003


On Oct 9, 2003, at 8:24 AM, Fabrice Popineau wrote:

>> Not always.  If there are files a.b and a.b.tex, then don't you want
>> tex to read a.b.tex?  That's why I programmed it that way in the first
>
> Maybe I'm wrong but I  think that over the time, many people tend to 
> stay
> away from multiple dots in path names, much more than say spaces in 
> path
> names ;-)

I suspect that multiple extensions are here to stay and in some areas 
even on the rise. Spaces in file names are problematic in shell-type 
environments and thus especially in those environments, it is more 
likely that people will keep away from them.

Besides, there is a nice hierarchy when using extensions. E.g. a file 
foo.tar.bz2 is a BZIP2 compresses TAR archive file.

> All in all, it is only a matter of deciding if you want extensions to 
> be
> meaningful or not. If they are (as the use  of .tex seems to imply) 
> they
> should always be, ie even for a.b, that is you want to feed tex with 
> the
> file 'a' of type 'b'. So multiple dots should  be prohibited. Ok, 
> that's
> only an  interpretation.
>
> MacIntosh  used to store the file  type in the  file itself  and in 
> this
> respect solved the problem.

They are moving away from this solution to file name extensions. E.g. 
.app for applications etc. The more complex file types have been 
replaced by file bundles (in fact directories that only look like files 
when inspected through the GUI). E.g. .app is a directory structure, 
not a file.

Hierarchy has the advantage here as well. A filoe type may be BZIP2 
compressed file, but .tar.bz2 shows you what has been compressed.

Actually, my installer does not rely on extensions, but uses the unix 
`file' command to try to find out what kind of files are in a package.

>> place, after years of horrible confusion about the rules.  In
>> retrospect, the whole idea of adding default suffixes is probably not
>> the right thing to do.
>
> Olaf suggested to make them configurable, which would help.
>
>> Anyway, forget that, it doesn't matter now.  What's matters is either
>> getting bibtex8 fixed or getting kpathsea fixed so it works again.  Or
>> dropping bibtex8, as Hans suggested.  I really don't care which,
>> myself, and I can't offer to do the programming :(.
>
> No opinion about it.
>
> -- 
> Fabrice
>
> _______________________________________________
> TeX Live mailing list
> http://tug.org/mailman/listinfo/tex-live



More information about the tex-live mailing list