[tex-k] -format/-progname

Paul Vojta vojta at Math.Berkeley.EDU
Wed Oct 29 04:05:46 CET 2003


> From: Reinhard Kotucha <reinhard.kotucha at web.de>
> Date: Wed, 29 Oct 2003 03:32:08 +0100
> To: Hans Hagen <pragma at wxs.nl>
> Subject: Re: [tex-k] -format/-progname
> Cc: tex-k at tug.org, Karl Berry <karl at freefriends.org>
> 
> >>>>> "Hans" == Hans Hagen <pragma at wxs.nl> writes:
> 
>     > I fully agree with you that -progname directive is indeed fuzzy
>     > esp in combination with "other text files" (btw, just 'other'
>     > would be more comfortable, esp since supporting "quoting" is
>     > painful in cross platform apps).
> 
>     > Maybe we need categories: 'programs and scripts'
>     > (pl,rb,py,java,exe,cmd,sh), documentation (pdf,ps,dvi,htm,html),
>     > fonts (pl,...)
> 
> As Akira Kakuto has confirmed my assumption that -progname just causes
> kpsewhich to search in $TEXMFS/<progname> I'd say that the
> documentation is not clear.
> 
>    kpsewhich -format='other text files' -progname=fonts readme.txt
> 
> finds a readme.txt in $TEXMFS/fonts.

I disagree.  The -progname argument affects how the kpse library interprets
the texmf.cnf file.  Setting -progname=tex, for example, causes kpsewhich
to accept lines of the form VAR.tex=value instead of ignoring them.

While it is true that some parts of the standard TDS tree are named after
programs (e.g., dvips), that fact is not reflected in the design of
kpsewhich; it is only borne out by the setup in texmf.cnf.

Clearly progname should suggest "program name".  There is no program
named "fonts" in TeX, nor any program named "doc".  There are programs
named "tex", "latex", "dvips", "xdvi", etc., and if you look in texmf.cnf,
you will see those program names appearing after environment variables
in lines of the form "<var>.<progname>=<value>".

> So there is no need for a table lookup.

[snip]

> I dislike table lookups, especially if the tables are compiled into
> programs.  It would be much nicer if a program would use an algorithm
> to do it's job rather than look into a table.  If there is a need for
> a table lookup, the table shouldn't be compiled into the program.

IMHO, kpse is already horribly over-engineered as it is.

--Paul Vojta, vojta at math.berkeley.edu


More information about the tex-k mailing list