Paul Eggert wrote: > diff --git a/m4/realloc.m4 b/m4/realloc.m4 > index 639d7a9efe..4a3f2ea8ae 100644 > --- a/m4/realloc.m4 > +++ b/m4/realloc.m4 > ... > @@ -70,19 +33,45 @@ AC_DEFUN([gl_FUNC_REALLOC_POSIX], > AC_DEFINE([NEED_SANITIZED_REALLOC], [1], > [Define to 1 if realloc should abort upon undefined behaviour.]) > else > - if test $REPLACE_MALLOC_FOR_MALLOC_POSIX = 1; then > - REPLACE_REALLOC_FOR_REALLOC_POSIX=1 > - fi > + REPLACE_REALLOC_FOR_REALLOC_POSIX=$REPLACE_MALLOC_FOR_MALLOC_POSIX > fi > ])
This change is not wrong. But it made the code more fragile: Its correctness now depends on the fact that gl_FUNC_REALLOC_POSIX is AC_REQUIREd before the assignments in line 78 and line 80. This is one more thing to watch out in future refactorings. I generally favour code that is a bit more verbose if it can avoid future bugs. Bruno
