Re: FYI: fix for ansi2knr and LIBOBJS

[Jim Meyering]

Alexandre Duret-Lutz <address@hidden> wrote:
> I'm checking this in on HEAD and branch-1-7.  This fixes a bug
> that affects any package using LIBOBJS and ansi2knr.
>
> For instance Coreutils 4.5.1's lib/Makefile.in contains
>
> chown_.c: lib/chown.c $(ANSI2KNR)
>         $(CPP) ... | $(ANSI2KNR) ...
>
> instead of
>
> chown_.c: chown.c $(ANSI2KNR)
>         $(CPP) ... | $(ANSI2KNR) ...
>
> (Jim, I'm surprised you haven't got any report about this.  There
> are many occurences of such bogus rules, even in src/.)
>
> 2002-12-01  Alexandre Duret-Lutz  <address@hidden>
>
>       * automake.in (lang_c_finish): Do not fill de_ansi_files
>       with subdirectory informations for %libsources files.
>       This revers part of a change from 2002-01-13.
>       * tests/ansi10.test: New file.
>       * tests/Makefile.am (TESTS): Add ansi10.test.

Hi, Alexandre!

Thanks for fixing that.
I think the reason is that very few people even try
to build with K&R compilers any more.

I will eventually remove all ansi2knr support,
since it's been some time since a K&R compiler could
build the *utils.  I think the first `offense' was due
to string concatenation in regex.c.

Jim