Paul Eggert <[EMAIL PROTECTED]> writes: > Simon Josefsson <[EMAIL PROTECTED]> writes: > >> Possibly the bug-workaround for unistd.h should only be enabled if the >> sysexits module is present, otherwise the unistd-module would always >> touch non-standardized namespaces which seems wrong. > > I agree. I'd rather not use the sysexits module, which means I'd > rather not have the unistd module messing with EX_OK (for example, > lib/euidaccess.c or similar code may need the system unistd.h EX_OK).
Right. >> e) Create a replacement <unistd.h> that does: >> >> #include </usr/include/unistd.h> >> #if @GNULIB_SYSEXITS@ >> # undef EX_OK >> #endif > > That doesn't look right, since code would have to include unistd.h > before sysexits.h. > > How about (b)? It seems the simplest. > >> (b) Create a replacement <sysexits.h> that does >> >> #include <unistd.h> >> #undef EX_OK >> #include </usr/include/sysexits.h> Yes, I think so too. That may cause problems if someone is using EX_OK without using the sysexits module with a buggy unistd, but they should really be using the sysexits module then. Most likely, using EX_OK without the sysexits module will break on some other platforms, and then they will start using that module. Perhaps a 'gnulib-lint' could be useful? The tool would search your code for non-portable function usages and suggest names of gnulib modules that you'll need. If it finds EX_OK in your code, it would suggest adding the sysexits module. /Simon
