>>>>> "Angus" == Angus Leeming <[EMAIL PROTECTED]> writes:
Angus> Let's address each issue as I post a chunk. However, I have a Angus> general question: Angus> os_win32.h is a kludge. It is performing the same rôle as Angus> config.h. I think that these things should go in config.h, Angus> which suggests a config/win32.m4 file. * Do you agree? * Is it Angus> possible to add tests to configure that effect only Win32 Angus> users. Ie, config.h won't be changed as I add tests except on Angus> Win32 machines? Another possibility is to put this in win32_defines.h, like we have support/os2_defines.h (that gets automatically included from config.h as needed). But I guess Lars wants to get rid of that. I guess we should consider the headers one by one, and try to isolate the places where they are used. The os-dependent stuff should be isolated in lyxlib.*, I guess. In any case, things like +#ifndef __MINGW32__ #include <sys/wait.h> +#endif should be replaced by +#ifndef HAVE_SYS_WAIT_H #include <sys/wait.h> +#endif and we should test for sys/wait.h in configure if we do not do it yet. But I think such headers should be included only in some specific os-dependent places. JMarc
