On Sat, 24 Nov 2001 23:29:34 +0100, abel deuring wrote: >> 1. sharp.c refuses to compile because of the last changes in CVS: >> >> sharp.c: In function `do_cancel': >> sharp.c:2262: storage size of `ds' isn't known >> sharp.c:2264: warning: implicit declaration of function `shmctl' >> >> I think that's because I compile without HAVE_SYS_SHM_H defined >> in sharp and nec backends) and so sys/shm.h is not included... >> >> (There is a sys/shm.h available for OS/2. I just renamed it some >> time ago to avoid fork() so it can't be found by configure)
>Oops. Did you use the CVS version of sharp.c or the one from the 1.0.6 >tarball? 1.0.6 tarball is OK, the problem was in CVS version. > >The CVS version contains an idiotic leftover from a test: > >#ifdef HAVE_SYS_SHM_H >#define USE_FORK >#endif >#define USE_FORK > >The last line is obviously nonsense... > >If the error occurs for the version from Sane 1.0.6, we'll need a better >way to make sure that USE_FORK is not defined for OS/2. For OS/2 it would be better if USE_FORK could be turned off independent from having HAVE_SYS_SHM_H defined or not. soemthing like #ifdef HAVE_SYS_SHM_H 'and not defined' HAVE_OS2_H >#define USE_FORK ... Sorry I don't know the right syntax. Franz
