Bruno Haible <[EMAIL PROTECTED]> writes: > Gnulib's purpose is to provide portability across the mainstream > systems, _regardless_ whether they implement the standards > correctly.
I agree with most of the other points you made, but I'd like to demur on this point. Gnulib is part of the GNU project, and the GNU project's goal is to develop a free operating system. The GNU coding standards make it clear that porting to incompatible proprietary operatings systems like MSDOS, VMS, etc., is low on our long list of things to do. Obviously some judgment is required, but in this particular case I see no real harm in assuming that malloc conforms to POSIX with respect to errno. If we don't attempt to work around the problem, MSDOS users might get a confusing diagnostic or other such behavior when they run out of memory. But that's not a big deal. If we upgrade the already-existing malloc wrapper to set errno to ENOMEM when malloc fails, and fix the m4 code to detect the MSDOS incompatibility with POSIX, that would be icing on the cake. This approach would be far preferable to modifying all our source code to worry about non-POSIX malloc implementations. Also, I note that in Visual C++ 2005, malloc sets errno to ENOMEM if memory allocation fails; see <http://msdn2.microsoft.com/library/6ewkz86d(en-us,vs.80).aspx>. So perhaps the simplest solution here is to assume POSIX behavior and wait for Microsoft users to upgrade themselves. _______________________________________________ bug-gnulib mailing list bug-gnulib@gnu.org http://lists.gnu.org/mailman/listinfo/bug-gnulib
