Re: teTeX and TeX Live interoperability
Ralf Stubner <ralf.stubner@web.de> wrote:
>> There might still be a
>> > problem with the order of trees: In upstream teTex, it is
>> >
>> > ...,!!$TEXMFMAIN,!!$TEXMFLOCAL,!!$TEXMFDIST
>> >
>> > But in order for TEXMFLOCAL to be able to shadow files installed with a
>> > Debian package, either Debian packages have to install in one of the
>> > TEXMFDIST trees, or we must keep on swapping MAIN and LOCAL.
>
> I would prefer the latter approach.
Does anybody know why TEXMFMAIN is before TEXMFLOCAL? I suspect the
reason is that in Thomas' standard setup, the contents of TEXMFMAIN are
what comes with tetex-src, or tetex-bin in Debian terms, and TEXMFMAIN's
precedence avoids to use wrong pool files that accidently get into
TEXMFLOCAL.
>> Several comments:
>>
>> (1) Order is important: which of teTeX and texlive do we wish to take
>> precedence? Most relevant is probably: which is more likely to be
>> more recent?
>
> In most cases, I would expect TeX Live to be more recent.
Thas also depends on whether Norbert is going to package snapshots into
unstable, and lets them migrate to testing, or not. For example, if
we'd had tex-live in spring 2004, and we had anticipated that sarge
would be released in the first half of 2005, then I would have uploaded
the teTeX-2.9x betas to unstable, and we'd had teTeX-3.0 in testing 10
days after it was released. If Norbert decides to only let released
texlive versions into testing, sarge would have shipped with tex-live
2004 (AFAIK from late summer 2004) and teTeX-3.0 from January 2005.
It also depends on the way texlive development works - are there
functional snapshots at all, a quarter of a year after the release?
>> (2) Use of TEXMFDIST
>>
>> (Incidentally, why have we got TEXMFDIST=$TEXMFMAIN in our default
>> texmf.cnf? That might cause two lookups if a file isn't in
>> TEXMFMAIN. Why not just comment it out everywhere it appears, and
>> list those variables which will need to be changed if TEXMFDIST is
>> set?)
>
> Good question.
I know why: Because I simply didn't know :-(. I didn't know whether
not defining TEXMFDIST at all wouldn't cause errors, and I never came to
testing it. Or maybe I even talked to Thomas and found that it would
cause problems?
Regards, Frank
--
Frank Küster
Inst. f. Biochemie der Univ. Zürich
Debian Developer
Reply to: