Re: teTeX and TeX Live interoperability
On Fri, Nov 04, 2005 at 09:34 +0100, Frank Küster wrote:
> Ralf Stubner <ralf.stubner@physik.uni-erlangen.de> wrote:
>
> > On Thu, Nov 03, 2005 at 14:27 +0100, Frank Küster wrote:
> >> In this setup, we still have the problem: Where can other TeX-related
> >> packages put their configuration files?
> >>
> >> - Either they install normal files into TEXMFMAIN and create symlinks to
> >> places in /etc/texmf/ (Not {TeXLive,teTeX}),
> >>
> >> - or they install normal files into TEXMFSITE *always* and put
> >> configuration files into /etc/texmf/texmf-site.
> >
> > Not sure. One problem is that TEXMFSYSCONFIG comes extremly early in the
> > search path in some situations. updmap-sys (and probably fmtutil-sys and
> > texconfig-sys) set TEXMFCONFIG equal to TEXMFSYSCONFIG.
>
> What could be the problem with that?
That is the second half of #334747 so to say. Imagine a package (or the
local admin) wants to shadow a config file in TEXMFSYSCONFIG. No
problem, just place an equally named file in TEXMFMAIN (or TEXMFLOCAL),
which comes before TEXMFSYSCONFIG in the search path. However, this is
not the case for updmap-sys, fmtutil-sys and texconfig-sys. These
programs set TEXMFCONFIG equal to TEXMFSYSCONFIG. TEXMFCONFIG is the
first TEXMF tree in the search path, so now files in TEXMFSYSCONFIG
are no longer shadowed by files TEXMFMAIN (or TEXMFLOCAL).
Fortunately, the *-sys programs don't read many configurations files,
and things like updmap.cfg are auto-generated anyway (but see #334747).
However, map files could be problematic. BTW, I am not sure why somebody
would want to shadow a map file, but then, I am don't understand why map
files are config files in the first place. ;-) Therefore it might be
that the dangers I am seeing are purely hypothetical.
cheerio
ralf
Reply to: