Hi Simon, > #elif 1 > # undef close > # define close close_used_without_requesting_gnulib_module_close > ... > Is it appropriate to use a hard error in this situation? If it was a > warning, it would be something that could be ignored.
Hmm, the presence of the 'sys_socket' module is an indicator that sockets are being used. But not a strong enough indicator, probably. How about keeping the error only if @UNISTD_H_HAVE_WINSOCK2_H@ && @GNULIB_SOCKET@ and instead use a link-time warning if @UNISTD_H_HAVE_WINSOCK2_H@ && [EMAIL PROTECTED]@ ? > Should the lseek-tests module depend on close? That feels wrong to me, > it doesn't use close in a non-portable way, and certainly not in the way > that the close module is intended to work around. I agree. The documentation of the 'close' module does not indicate that 'lseek-tests' needs it. Hence the compilation should also work without the 'close' module. Bruno
