Bugzilla Bug 124511

I'll like to use this bug as a meta-bug that keeps track of all the issues that
will eventually lead to tetex-3.0_p1 is marked stable.
(This will result in many tetex-2 only bugs going away :-)

This bug should depend on all the bugs that causes tetex-3.0_p1 not to be a
candidate for stable.
I have added some bugs already, fell free to add/remove bugs you think this bug
should/should not depend on.

We should also discuss other issues, like the problem with beamer and pgf.
There are are old versions in the texmf source, but we cannot add new versions
due to file collisions.
A imho good solution to this problem would be to add a new dist texmf tree with
higher priority than the standard tree and then have the latex-package.eclass
install packages to that tree.
This could waste some disk space for users, becuase they could end up with
duplicated packages in the two trees, but dev-tex packages are normaly small,
so I think the majority of users wont see it as a problem(?).

When we have taken care of all the problems, archs should be added so they
could mark it stable.

Bug 68878 does not apply to tetex-3. According to upstream they have fixed it
in tetex-3 (I cannot reproduce it with tetex-3 either).

That on the other hand does not mean that tetex-3 passes make test. It still
fails on at least on sub-target (i'm currently trying to fix it).

I'm currently trying to implement the same structure as perl uses for perl
modules (take a look in virtual/perl-*) for tex in my overlay.

It will give a slight overhead for package maintainers but will avoid wasting
space on users systems.

Any comments?!?

I have removed dvipng from the latest tetex ebuild since there is a file
collision between tetex and dvipng (dvipng is shipped with tetex and comes in
its own package app-text/dvipng).
So we probably need to have a stable version of dvipng by the time we mark the
version of tetex without dvipng stable (so users with stable archs can have
dvipng installed).

According to

http://news.gmane.org/gmane.comp.tex.tetex.general

there won't be a next release of teTeX. tetex-3.0_p1 consists of a beta version
of teTeX with a not up to date texmf-tree. As far as I know Debian is switching
over to TeX Live. What is the future of TeX in gentoo?

I think this should be cleared out before marking tetex-3.0_p1 stable.

The fact that Thomas stopped tetex has nothing to do with tetex-3 going stable.
But for your information my (our?) plan is the following:
1. Mark tetex-3.0_p1 stable.
2. Make all the dev-tex packages use the new texmf-site directory, so that we
can remove all those stupid blocks. This is not as simple as it sound, since we
need to think about how to resolve deps between packages which ships with the
texmf-tree and those installed in separate packages (it looks like new-style
virtuals are a good candidate, but apparently they are not the holy grail, see
-dev mailinglist).
3. Create a ~arch version of texlive, and have ~arch users switch.
In that process we need to figure out from where we are going to get an update
texmf-tree, since Thomas won't maintain one. (suggestions are more than
welcome).
4. Mark texlive stable and kick tetex from the tree.

The above process can (probably will) take at least a year (unless text-markup
suddenly gets flooded by new devs).
But I'm working as fast as I possible can!

An additional idea I have is to make our texlive package much more modular (it
seems to be the next big thing :). At least split it into a texlive and
texmf-tree package. But possible also into a kpathsea and if useful also
several other packages. It seems stupid that a package which requires the
kpathsea lib must install a whole tex distribution.

I'll probably make an announcment on the -dev mailinglist about the above.

Thanks for the information! As for a recent texmf-tree wouldn't it be possible
to use or change the TeX Live install-pkg.sh script to extract the desired
collections and/or packages into an archive?

E.g. "sh install-pkg.sh --collection=xxx --archive=texmf.tar" produces such an
archive. What remains to do is writing an adequate collection-file (xxx in the
above command).

As for marking tetex-3.0.p1 stable: Im am using tetex-3.0_p1 since it is in the
portage tree without any problems.

tetex used to have both the mips and sh keyword, but they were both dropped:

mips: Was dropped due to bug 87627 (a dep problem which has been fixed)
sh: Was dropped due to unknown reasons when going from tetex-2 to tetex-3 (I
wasn't on the team at that point and I can't find any details about why it was
dropped in the ChangeLog or any other place for that matter).

Both arch teams could you please consider marking tetex-3.0_p1-r3 with a
unstable keyword? (I'll get back to you about stabelizing when the unstable
keywords has been in the tree for a month).

> Both arch teams could you please consider marking tetex-3.0_p1-r3 with a
> unstable keyword? (I'll get back to you about stabelizing when the unstable
> keywords has been in the tree for a month).
> 

Is there actually a reason to have mips keywords other than, "because my ebuild
is shiny and speshul and I want you to!" ?  We clearly have no user demand for
this package on mips, otherwise we would have seen complaints and/or bug
reports from users.

(In reply to comment #8)
> Is there actually a reason to have mips keywords other than, "because my ebuild
> is shiny and speshul and I want you to!" ?  We clearly have no user demand for
> this package on mips, otherwise we would have seen complaints and/or bug
> reports from users.

Because the keyword used to be there. There is a stable mips version
(2.0.2-r5), and I want to kick tetex-2.* when tetex-3 is stable. I can't do
that (in good faith) unless you eventually keyword tetex-3 or remove the mips
keyword.
So if you don't think a mips keyword is need _at all_, then feel free to kick
the keyword from _all_ tetex ebuilds.

Any news on the issue above from mips and/or sh?!?

mips: Please get your act together and take a stand on this one! Do you want
the mips keyword or not?

Is there any reason why tetex-3 can't be stable on x86 now? I've been using it
for over 6 months with no problems, and this bug doesn't seem to depend on any
others now...

(In reply to comment #12)
> Is there any reason why tetex-3 can't be stable on x86 now? I've been using it
> for over 6 months with no problems, and this bug doesn't seem to depend on any
> others now...

Acutally there has been some bug reports recently, which I haven't had the time
to look through, which might block this (I'm currently dev-away, due to
finishing my thesis and my wife giving birth).
I hope to be back at the end of august and then I'll have a look at it!.

TeTeX 3.0_p1-r3 is stable now, but I can't find the hyphenation pattern for
Italian. As far as I can understand when I run texconfig, there are only
patterns for english, mongolian, spanish, russian, greek, inuit, dumylang and
nohyphenation.

Sorry, I've reemerged tetex-3.0_p1-r3. The file language.dat was complete but
babel loads all of the languages.
Moreover, latex creates pdf files, not dvi ones.

alpha stabilized -r3

Closing, mips apparently doesn't want this package.