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

Re: lenny release goals and the tetex->texlive transition



Hi,

here comes the answer for the rest of your mail.

Norbert Preining <preining@logic.at> wrote:

> Hi all!
>
> On Die, 12 Dez 2006, Frank Küster wrote:
>> all:  Make sure that all packages work with texlive.  And actually it
>> wouldn't hurt to get some of those fixes into etch...
>
> Yup. This probably means doing some NMU, as there are packages with bugs
> open since some time and no reaction, although there have been new
> versions on a regular basis (ipe, I even prepared a new source package,
> texmacs, ..)

Hm, I'm not sure we can NMU packages targetted at etch.  It's not a RC
bug to not support TeXlive.  Maybe we should instead provide empty
"tetex" packages prepared with equivs for etch users, somewhere on
tug.org?

>> ***** Transition from teTeX to TeXlive *****
>> This can be divided into a couple of sub-issues:
>>     * Build libkpathsea4 from either TeXlive, or a separate pacakge
>
> This is an interesting question. I have NEVER checked whether it would
> be possible to build libkpathsea dyn version from the TeX Live sources.
> And to be honest, I am feeling myself not very able to to check for this
> in the hell of configure/auto*/MAkefile.in hell.

I think in tetex it just means to pass --enable-shared to configure, and
maybe some tweaks in the install targets.   I don't think it would be
hard with texlive.  But still, a separate package has its merits.

>>     * Make sure that TeXlive is always the primary choice
>
> This needs a long way to go, as many packages already have alternative
> deps on texlive, but as second option.

Yes - this becomes unnecessary in case we drop the tetex packages.  But
if we don't do that, there's no way to ensure that buildds consistently
install texlive, I fear.

> Hahaha... "Completely" I am not even DREAMING of this. 

Err, yes, "Completely" was copied from the etch release goals.  Utopia.

> But in fact what
> I did before leaving was something more desperate: Getting AUTOMATIC
> update checking from CTAN to the TL svn repository working: This looks
> like:
> - Check for update available
> - check whether current default routine works
> 	. if not adapt it, if yes be happy
> - check the license
> - add it to the autocheck list
> There is a nice document written on this, if someone (the students!?!?!)
> want to get their hand on:
> - perl programming
> - license checking
> - subversion usage
> this is the *perfect* option, as all of this is needed.

I'm not sure whether they will find this invitation - can you please
write them and Rogério a separate e-mail?

>>     * Document the packaging better, so that the unititiated are able to build
>>       a package from SVN.
>
> Hmm, of course it needs better description, but I think the README
> already describes quite good how to make a source package.

No, not really - I always need to do trial-and-error if the last
tpm2deb.pl invocation is no longer in the bash history.  The needed
options are not documented (I think it's --master and --oldsource), and
the targets aren't described clearly, too (what is the make_deb target
for?  It simply does nothing IIRC).

>> * there will be some mess with (build-)dependencies where 'texlive' is
>>   not enough; I hope it is closer to 2. than to 1., but of course I am
>>   not sure
>
> I still hope that this will not be too bad. 
>
> As I said, I vote for the transitional packages, and that move should be
> done immediately after etch released.

Before we do that, we should check

- whether everything in tetex-base + tetex-bin is included in the
  texlive metapackage.  I can imagine that some style has been installed
  in -base, but is in tl-latex-extra.  If we find this, we should try to
  get an idea whether any build-depending package might use this.

- whether we can specify a decent list of texlive packages on which the
  tetex-extra transitional package depends on.  If this ends up to be
  2/3 of texlive-full, it's not practical.

Regards, Frank
-- 
Dr. Frank Küster
Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich
Debian Developer (teTeX/TeXLive)



Reply to: