[tex-k] Twiki is broken; sorry . . . let's use tex-k list if that's okay.

Fabrice Popineau Fabrice.Popineau@supelec.fr
Tue, 06 Nov 2001 07:49:59 +0100


* Thomas Esser <te@informatik.uni-hannover.de> writes:

>> I'm very sorry and very red-faced and quite disappointed.
> Argh... Can't you really find nothing of it in the rcs files? Does
> anybody have my text in his browser history (I do not!). It really
> took quite some time to write that down...

Do you mean this :

Tom E.: I think that a few points need clarification. 

* web2c and kpathsea are maintained by Olaf Weber although some of his
  work was never published independendly of teTeX or TeXLive

* The parts of teTeX: the first part is web2c and kpathsea which I try
  to include as  literally as possible.  The  second part of teTeX are
  drivers,  pdfTeX, Omega, e-TeX.  Not   all, but some  of these  have
  their own maintainer   who usually  makes  sure  that  his stuff  is
  somewhat "in sync" with Olaf's  work. Again, I  try to include  this
  kind of things as unchanged as possible.  The third part is software
  that has  a  maintainer  but  no direct  relationship  to TeX,  e.g.
  ncurses, libwww, zlib,  ... For teTeX, I have  made them work in one
  common source tree, i.e.  I have  set up the  autoconf stuff for the
  libraries,  set  up  the  paths  etc.  etc.    My changes  regarding
  Makefiles, configure scripts etc.  usually have  been adopted by the
  maintainers  of the first two  categories. The  fourth part are some
  programs (all scripts)  which I maintain myself: fmtutil, texconfig,
  texdoc and a few others.

* texk is  a collection  of web2c,  kpathsea  and a few   drivers. The
  drivers are being taken from teTeX at the moment, but one cannot say
  that     web2c/kpathsea is a     subset   of teTeX.  Development  of
  web2c/kpathsea is done by Olaf Weber, not by me.

* TeXLive is a complicated thing. I think that Sebastian et. al.  have
  collected patches /  improvements  for  all  kind of  packages  that
  TeXLive includes. Its big advantage at the moment is that its source
  tree is open to several developers.

The most important point  of driver development in  my opinion is that
it   should not depend  that      much on web2c/kpathsea,  teTeX   and
TeXLive. Driver developers should just maintain their code and nothing
else. One solution is that we all go into one  common source tree, but
I  think that there is  a better  solution which  is  based on a small
kpathsea package where these drivers could be based on. Such a package
would be needed to two things:

* it can provide kpathsea as a shared library where drivers could just
  link against

* it can provide  a source tree where the  driver can  be "plugged in"
  (for people who want to compile the driver with  a static version of
  kpathsea). Ideally,  the    driver itself would   make  just   a few
  assumptions about the actual  implementation of the kpathsea package
  (and thus be robust against changes in the kpathsea package).


Fabrice