Planning for TL5

Arthur Ogawa ogawa@teleport.com
Mon, 23 Aug 1999 22:26:56 -0700


I had a very productive talk with Sebastian and Fabrice at TUG 1999. As
an outcome of these discussions, I would like to present this list of
talking points.

1. I have been quite exercised that with the generation of TeX Live 5,
we would be in a situation where, albeit we had successfully reached the goal
of allowing anyone to make TeX Live 5 CD and sell them at any price, we would
have in effect given away the candy store because CheapBytes would now be able
to manufacture and sell the CD for the canonical $2.

2. I identified two things that remain substantial added value and which can
be the basis for an incentive for TeX users to join TUG or a LUG, namely
documentation and enduser support. It is stipulated that the documentation
must be very much improved in order to serve as an
installation guide for new TeX users.

3. Sebastian stated that he would remove the documentation from the CD.

4. I talked up the possibility of developing the documentation through a
collaboration of LUGs who would each publish it in thier respective
national language. Volker Schaa is sympathetic to the idea. Fabrice, perhaps
you know of Francophones who might be willing to participate.

5. Sebastian talked about how he generates the texmf tree for TL, using
a Perl script he calls the sausage machine. Certain packages are unsuited to
this approach, particularly ones that contain files whose processing is not obvious.

6. Sebastian outlined the process of generating the TeX Live CD; there are
four parts to the work, namely binaries, executables, documentation, and
another part I cannot recall right now.

7. We agreed that it would be ideal if the submitter of a package would be so
kind as to provide build instructions for the package.

8. We also discussed that it would be in principle possible for an author to
prebuild all of the files needed at run time, albeit this would load up CTAN more.

9. We agreed that it would be desirable for the author to provide information
about the package, possibly in the form of an XML document that would be the
basis for an entry in Graham's database.

10. We agreed that it would be very good to create a "TeX Live" host for
building up the TL texmf tree. This host would be accessible by those working
on the CD, so it would facilitate our sharing out the labor. It might be
tug.org, but it might usefully be a CTAN host itself, such as CTAN.tug.org.

11. We discussed creating a system that would update a user's texmf tree over
the internet. This updater would presumably be based on the TL host mentioned above.

12. We discussed the possibility of rewriting the TL installers all in Perl,
so that they would run on all hosts, not just UNIX.


That's all I can remember of the conversations, so please feel free to fill in
the gaps. Note that I am by no means reporting our having made hard and fast
decisions: the above are only talking points.