Re: gnulib-tests/test-stdbool.c vs gcc2.95/OpenBSD

[Bruno Haible]

Hello Christian,

Christian Weisgerber wrote:
> Here's another issue I noticed while running the gettext-0.18.1
> regression tests on OpenBSD/sparc (gcc2.95):
> 
> In gettext-tools/gnulib-tests...
> 
> source='test-stdbool.c' object='test-stdbool.o' libtool=no  DEPDIR=.deps 
> depmode=gcc /bin/sh ../../build-aux/depcomp  cc -DHAVE_CONFIG_H -I. -I..  
> -DIN_GETTEXT_TOOLS_GNULIB_TESTS=1 -I. -I. -I..  -I./.. -I../gnulib-lib 
> -I./../gnulib-lib  -I../intl  -I/usr/local/include  -O2 -pipe -c -o 
> test-stdbool.o test-stdbool.c
> test-stdbool.c:36: syntax error before string constant
> 
> Specifically, it's  "error: true is not 1".
> 
> The relevant part of OpenBSD's stdbool.h is this:
> ------->
> /* `_Bool' type must promote to `int' or `unsigned int'. */
> typedef enum {
>         false = 0,
>         true = 1
> } _Bool;
> 
> /* And those constants must also be available as macros. */
> #define false   false
> #define true    true
> <-------
> http://www.openbsd.org/cgi-bin/cvsweb/src/include/stdbool.h?rev=1.4
> 
> So, true *is* 1, but this isn't visible at the preprocessor level.

Hmm. How come that stdbool.m4 did not catch this? Can you please look
for the values of STDBOOL_H and HAVE__BOOL in gettext-tools/config.status?

$ grep STDBOOL_H gettext-tools/config.status
$ grep HAVE__BOOL gettext-tools/config.status

Bruno