[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#173872: tetex-bin: spurious dependency on perl-tk



On Wed, Dec 25, 2002 at 11:47:02AM +0900, Atsuhito Kohda wrote:
> > > > we had a tex system like this up to Debian 1.2 containing up to 20
> > > > packages, but because of the complex dependencies it was not really
> > > > working this way. So the decission was, to adopt TeTeX because of its
> > > > good design and stability. If we again try to separate it in small
> > > > pieces I fear we get the same mess as before. So my answer is, rather
> > > > don't do it. 
> > 
> > That's all fine and well, but it was written in 1999, and is referring to
> > something that happened in 1996.
> 
> No, it is not old story.  It is a main spirit or good design
> or excellent feature of teTeX to provide integrated and consistent
> TeX system.  This is definitely true even today.

And that's completely orthogonal to the concept of splitting out different
parts into different packages.

> TeX system consists of binaries and their supporting stuffs, 
> like configuration files and/or various kind of fonts (TFM, 
> PK, Type1, VF etc.) and all of them should be consistent.

So use meta packages, versions in dependencies, carefully organize virtual
packages etc.

There's ample proof in the distribution that complex packages can indeed be
done. I cannot assert that teTeX isn't in fact harder than anything else
we've ever seen, but then, I don't see bugs filed against dpkg about these
grave deficiencies in the packaging system that prevent teTeX from being
packaged properly.

> Further, there are even requests to add some components to 
> tetex packages (cf. #31897, #99154, #105333, #129758 and we 
> have gotten requests in debian-tetex-maint ML directly).

But like I just said, that's not related to how these things need to be
packaged. The very word you use, "component", describes this nicely: the
system may not be complete without any one of these, but they are _not_
indivisible from the whole!

> That is, this issue heavily depends on users and it needs
> very grave/common/reasonable reason (instead of unnecessary
> nor small disk space) to split tetex into a few componets.

It is disturbing to see a packager not concerened at all with irrational use
of disk space. It is downright saddening to see so little sympathy for those
users that don't have endless amounts of disk space that they are discarded
as uncommon and unreasonable.

-- 
     2. That which causes joy or happiness.



Reply to: