Building guile on MSYS/MinGW - issue with pthread use

[Gareth Edwards]

Hi there,

I've run into a problem building guile using the MinGW toolchain under MSYS.

address@hidden /c/gareth/geda/deps/guile-1.8.7/libguile
$ gcc -DHAVE_CONFIG_H -I.. -I/usr/local/include -g -O2 -Wall -Wmissing-prototyp
es -Werror -MT libguile_la-threads.lo -MD -MP -MF .deps/libguile_la-threads.Tpo
 -c threads.c -DDLL_EXPORT -DPIC -o .libs/libguile_la-threads.o
threads.c: In function 'scm_timed_wait_condition_variable':
threads.c:1312: error: storage size of 'waittime' isn't known
cc1.exe: warnings being treated as errors
threads.c:1312: error: unused variable 'waittime'

I believe it's the same issue as mentioned here:

http://lists.gnu.org/archive/html/guile-devel/2008-07/msg00052.html

but it doesn't look like it was resolved at that point. Hopefully I
can explain what I think is going on. I think the fundamental issue is
that pthread.h and guile are using the same #define name.

pthread.h defines timespec like this:

#ifndef HAVE_STRUCT_TIMESPEC
#define HAVE_STRUCT_TIMESPEC 1
struct timespec {
        long tv_sec;
        long tv_nsec;
};
#endif /* HAVE_STRUCT_TIMESPEC */

So, presumably, configure tests for timespec, it doesn't have
HAVE_STRUCT_TIMESPEC set. Test passes, and then config.h then gets

/* Define this if your system defines struct timespec via either <time.h> or
   <pthread.h>. */
#define HAVE_STRUCT_TIMESPEC 1

But now, when config.h is included before pthread.h, this has the
effect of hiding this declaration in the pthread.h file, so the guile
build breaks.

Cheers
Gareth