Re: Too many conflicts? (tetex vs. texlive)
Florent Rougon <f.rougon@free.fr> wrote:
> Norbert Preining <preining@logic.at> wrote:
>
>> This sounds quite reasonable. Only depending on tex-common, and calling
>> mktexlsr and updmap-sys only if tehy are present.
>
> I'm *almost* convinced. :-) But...
> I'd like to hear Frank's opinion about that.
Uhm, I'm overwhelmed by the amount of mails in the list this morning.
Fortunately I do have time to read them, just run to the telefone every
30 minutes to check wether this guy is finally available...
But I might not have understood everything, and might miss some
important considerations.
Generally, I think that we should try to taylor the packages in a way
that
- only stuff that needs the "real" programs from tetex-bin (or its
texlive counterpart), like (pdf)tex, xdvi, dvips, need to depend on
them
- while others can get along with tex-common.
I think that this is already true for font packages, but I have never
tested it or even thought about it, except this morning.
> Before tetex-bin is configured, texmf.cnf might be broken or incomplete.
> updmap-sys creates files in $VARTEXMF/fonts/map/pdftex/ for instance,
> and that might change when a package such as tetex-bin is configured.
Since the basic components of texmf.d are in tex-common I don't see this
problem. Even if it happens that tetex-bin needs a change, this change
should be coordinated with Norbert and made in tex-common (and in the
Policy).
> But there _are_ corner cases, and I cannot assume without a word from
> the teTeX maintainers that running these scripts before tetex-bin is
> configured is supposed to work in all cases.
In an other mail, Norbert suggested to move the scripts that are common
to tex-live upstream and teTeX upstream, like updmap and mktexlsr, to
tex-common. The benefits would be obvious - most notably we would only
need one set of patches to them. On the other hand, there are also a
couple of disadvantages:
- When the scripts change incompatibly (e.g. like the "recent" change to
updmap not to silently skip nonexistent map files), uploads of teTeX
and tex-live must be coordinated.
- Upstream code gets into a Debian-native package.
If we do move the scripts to tex-common, we should at least put an
unchanged upstream version (either teTeX or texlive) into it, together
with Debian-specific patches.
Regards, Frank
--
Frank Küster
Inst. f. Biochemie der Univ. Zürich
Debian Developer
Reply to: