Eric Blake <[EMAIL PROTECTED]> wrote: > According to Bruno Haible on 1/10/2008 7:24 PM: > | Thanks for these. So the problem is that if __STRICT_ANSI__ is set, > | __attribute__(...) gets defined to empty. In some tests with GCC versions > | starting from 2.95.3, I found that __attribute__ works also with > | __STRICT_ANSI__, therefore there is no need to define it away. > > Seems fine to me, since the compilation failure was detected by my edits > to string.in.h, and I was just copying what was done elsewhere. Does > anyone know why we started checking __STRICT_ANSI__ in the first place?
That was necessary when -ansi would make gcc fail to recognize __attribute__. But that was long ago... maybe around 1995. Here's a clue: http://git.sv.gnu.org/gitweb/?p=gnulib.git;a=commitdiff;h=2917e4a652abe739fc45
