questions regarding patch-tmp

[frank@kuesterei.ch (Frank Küster)]

Hi all,

during the weekend, I had a look at the new upstream beta code and at
the patches we apply. A couple of questions arose, this mail contains
the questions for patch-tmp.

* The patch for resident.c failed. I don't really know C, but it looks
  like as if upstream has yet incorporated our fix, just with a
  different syntax. This is the old version we want to patch:

              ps->ysize = vsiz ;
-             ps->specdat = nextstring++ ;
              canaddtopaper = 1 ;

This is what we want:

              ps->ysize = vsiz ;
+             *(ps->specdat = nextstring++) = '\0' ;
              canaddtopaper = 1 ;

And this is what upstream has now:

            ps->ysize = vsiz ;
+           ps->specdat = nextstring++ ;
+           *(ps->specdat) = 0 ;
            canaddtopaper = 1 ;

It's o.k. to just keep it like that, right?

* regarding mktexlsr:

The ls_R_magic test is still not skipped if the ls-R file exists but has
size zero. I don't know anything about the history of this patch -
hasn't it been submitted to upstream? Or is it only a Debian problem
that somebody frequently touches empty ls-R files - then we should move
it to patch-deb, shouldn't we?

* currently we patch texk/klibtool, which is a home-tailored libtool
  version by Olaf Weber. Now upstream seems to use standard
  libtool. There is no /texk/klibtook any more, but there's
  texk/libtool.m4. I would be grateful if somebody who knows libtool
  would check the three versions (old upstream, debian patched, new
  upstream) and give me a hint.

* texmf.in: Why do we allow searches for files on disk instead of
  restricting searches to ls-R, as in upstream? This would speed up
  searching, and properly installed packages should regenerate ls-R
  anyway.

* any idea why we patch texk/xdvik/xdvi-sh, when we really use the perl
  script?

Thanks in advance, Frank
-- 
Frank Küster, Biozentrum der Univ. Basel
Abt. Biophysikalische Chemie