PING! Four unreviewed makefile and doc patches

[Kelley Cook]

PR/13820 "make dvi" crashes when configured from a relative source directory
http://gcc.gnu.org/ml/gcc-patches/2004-01/msg02349.html

This is a regression mistakenly introduced when *.dvi were moved to 
being built in the build directory.  It occurs due to a limitation with 
the tex2dvi included with Makeinfo 4.2.  The patch is a straightforward 
fix to use absolute path names instead of relative when calling texi2dvi.
------
Document new makefile targets and new lang-hooks
http://gcc.gnu.org/ml/gcc-patches/2004-01/msg02460.html

A doco patch as requested by Mr. Myers after I installed the patch to 
re-allow tarballs to be built without flex, bison or makeinfo.
------
Re: Document autoconf and automake requirements. (v2)
http://gcc.gnu.org/ml/gcc-patches/2004-01/msg00692.html

Another doco patch requested by Mr. Myers as followup to a patch which 
somewhat unified the required autofriends versions.  This version of the 
patch was the followup which should encompass the comments made after 
the generally accepted first version of the patch, but it was never 
explicitly approved.
------
Candidate for PR/13485 Generated files no longer get messages extracted
http://gcc.gnu.org/ml/gcc-patches/2004-01/msg02895.html

Without this patch, generated programs not in CVS such as c-parse.c do 
not get submitted to the translation project.  This is the only one that 
I think should be slightly controversial, since it will write to the 
newly instituted read-only source directory on the occasions that one 
wishes to regenerate gcc.pot.  As mentioned, I think that this fact 
could just be documented to an exception to the (currently undocumented) 
read-only source directory rule.