Re: CVS-19990122: PATCH for texinfo

[Jeffrey A Law <law at cygnus dot com>]

  In message <Pine.GSO.4.05.9901272314480.35-100000@markab.dbai.tuwien.ac.at>yo
u write:
  > In general, this texinfo stuff seems to cause quite some problems,
  > so it may be useful to consider the various options we have:
  > 
  >  1. Remove texinfo from the egcs tree and require a current version of
  >     texinfo to be installed.
  >  
  >     This will solve all problems of this kind, however it'll make egcs
  >     more "expensive". In addition we will have to check that the installed
  >     version is sufficiently up-to-date.
  > 
  >  2. Remove texinfo from the egcs tree and install the documentation only
  >     if texinfo is found.
  > 
  >  3. Keep texinfo and try to fix problems. That means, for example, that we
  >     will have to keep the util subdirectory in order to get install-info
  >     and install your patch (or something similar).
  > 
  >  4. Remove texinfo from the egcs tree and require it for CVS users, but
  >     deliver pre-compiled *.info files in snapshots and releases.
  > 
  >     (That won't solve the need for install-info, though -- in the absence
  >     of an installed version of install-info -- we could simply copy the
  >     *.info files.)
I would lean towards #4.  With the caveat that we don't bother with
install-info.  I just don't see the benefits outweigh the need to add more
complication to the makefile.

One thing I would like to look at between egcs-1.2 and egcs-1.3 is to greatly
simplify gcc/Makefile.  IMHO, it's not managable in its current state.  Many
things are obsolete.  Others can be made obsolete by updating some targets.

Once we clean up gcc/Makefile we should proceed up a level and look at the
toplevel configury and Makefiles.

I guess in general, I find that we've got a mess of code, some of which is
merely working around problems with work arounds of a bad design.

jeff