Re: Specifying TeX dependencies
Ralf Stubner <ralf.stubner@physik.uni-erlangen.de> wrote:
> Frank Küster wrote:
>
>> This puts the burden of ensuring that the functionality continues to be
>> available on the maintainers of the depended-on package, i.e. on the
>> people who know best.
>
> That makes sense for something like a library. But for a TeX
> distribution 'functionality' is such a wide field, I am not sure how one
> could ensure it.
Well, we can at least ensure availability of files and executables;
whether a new version of a LaTeX package still produces the same output
is of course beyond our power.
>> On the other hand, it makes it easier to
>> rearrange functionality in the packages that are only indirectly
>> depenended upon - we can e.g. take things out of tetex-base into
>> tetex-extra without bothering about depends, as long as there is a
>> tetex-bin package that pulls them all in (see
>> http://pkg-tetex.alioth.debian.org/splitting.html).
>
> Do you really think this makes it easier? Right now tetex-extra depends
> on tetex-bin which in turn depends on tetex-base. If we move some
> package from base to extra, it won't be installed if people depend on
> tetex-bin, only. Does tetex-bin then have to depend on tetex-extra?
If we do the splitting as discussed, we'll end up with three tetex-bin-*
packages - and I'd suggest a forth, tetex-bin, depending on all of them.
And this would then also pull in tetex-extra.
> Or should tetex-bin just depend on the to be created packages
> tetex-bin-core, tetex-bin-x11, and tetex-bin-extra which in turn bring
> in both tetex-base and tetex-extra. So in order to reduce the stuff for
> (build)dependencies people would have to change from tetex-bin to
> tetex-bin-core, right?
Yes, something like this was on my mind.
Regards, Frank
--
Frank Küster
Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich
Debian Developer (teTeX)
Reply to: