Bruno Haible wrote: > The bug Richard Lloyd reported was caused by the use of the "old" > idioms for function replacement, which don't allow a "#undef" for > reverting to the system's original definition. But such a "#undef" > is needed when lib/setenv.c is compiled as part of the relocwrapper > programs. > > As indicated in > <http://lists.gnu.org/archive/html/bug-gnulib/2010-06/msg00145.html> > most of the AC_FUNC_* macros that require replacement code will soon > defer to gnulib. Here we have the same situation for AC_FUNC_MALLOC and > AC_FUNC_REALLOC.
Thanks for diagnosing and fixing that. > Here are 3 patches to drop the uses of AC_FUNC_MALLOC and AC_FUNC_REALLOC. > In this proposal instead we would use _AC_FUNC_MALLOC_IF and > _AC_FUNC_REALLOC_IF, which are defined in Autoconf but undocumented. > I'm not sure which is better: using these undocumented macros, or > copying their code from Autoconf? (For calloc, gnulib already has its > own copy of _AC_FUNC_CALLOC_IF.) > > OK to apply? Looks good to me and passed this smoke test: ./gnulib-tool --create-testdir --with-tests --test malloc realloc calloc so yes. Richard, thanks again for the non-gcc testing and report. We don't get much of that, these days.
