Re: problems in stdbool.m4

[Bruno Haible]

Stepan Kasal wrote:
> > > m4_ifndef([AC_HEADER_STDBOOL],
> > > AC_DEFUN([AC_HEADER_STDBOOL],
> >
> > Nope, this would be a regression.
>
> Oh, I understand now. But I don't think gnulib should redefine documented
> Autoconf macros.  What about renaming it to gl_HEADER_STDBOOL ?

A guiding principle for gnulib is to write code in a way that abstracts
away other software's bugs. We write

  #include <getopt.h>     when /usr/include/getopt.h does not do what
                          we want

  bool var;               although gcc-2.95 didn't have a working 'bool'

  AC_HEADER_STDBOOL       because the test in autoconf-2.59 is buggy.

The idea behind this is
  1) to simplify our code,
  2) to make pressure on the other software maintainers, showing them how
     it would be correct,
  3) We want to minimize the work we have to do when a software fixes its
     bugs. (We have already enough work detecting the bugs...)

We do override include files by using our own if the system's one is defective.
We do #define bool to hide the one provided by the compiler.
We do override an autoconf macro if it is buggy.
Doing this is more reliable that using our replacement under a different
name, say <gl_getopt.h> or gl_bool or gl_HEADER_STDBOOL.

The hope is that someday there will be an autoconf release that fixes that
bug in AC_HEADER_STDBOOL.

Bruno