Re: Interference between two installations of gnulib

[Paul Eggert]

On 4/11/21 12:32 PM, Bruno Haible wrote:

> Having two separate configure scripts is undesirable, because that would
> increase the total configure time of the package (and people have already
> complained about the configure time).
>
> The same problem is seen also in GNU gettext and in GNU poke. I'm fixing it
> through the attached patch set.

Unfortunately this broke getopt-gnu on Solaris 10, because it relies on 
this line in modules/unistd:

	      -e 
's/@''GNULIB_UNISTD_H_GETOPT''@/0$(GNULIB_${gl_include_guard_prefix}_UNISTD_H_GETOPT)/g' 
\

and substitution went awry in this line, resulting in:

              -e 
's/@''GNULIB_UNISTD_H_GETOPT''@/0$(GL_GNULIB_GL_UNISTD_H_GETOPT)/g' \

This caused 'make check' to fail because test-getopt-gnu.c was testing 
Solaris getopt rather than the Gnulib substitute.

I discovered the problem after updating GNU grep to use the latest 
Gnulib, and after someone else's buildbots failed on Solaris 10 with 
'grep' and I got notified of their failures:

https://buildfarm.opencsw.org/buildbot/builders/ggrep-solaris10-i386/builds/227
https://buildfarm.opencsw.org/buildbot/builders/ggrep-solaris10-sparc/builds/229

I installed the attached patch to work around the problem. The Python 
patch is obviously a hack (and I haven't tested it); suggestions for 
improvements are welcome.

I must say that I am starting to reach my limits in debugging this sort 
of thing. We have quite a pyramid of hacks here, involving more than 
just the usual multilevel combination of make, m4, sh, and sed along 
with Git submodules etc., and now also a duplicate Python implementation 
that I don't offhand know how to test. And 'gnulib-tool', 'configure' 
and 'make check' are so verrryy slow; to find the commit that caused the 
problem, I had to run 'git bisect' overnight because my circa-2005 
Solaris 10 sparc machine is not as fast as modern machines. This is a 
long way from my traditional way of developing where I edited a Makefile 
and typed 'make' (and that was already too slow!).

I don't have a solution to this problem, and to some extent am just venting.

0001-getopt-gnu-port-back-to-Solaris-10.patch
Description: Text Data