On Tuesday 07 June 2011 11:40:08 Cactus wrote: > On Jun 7, 10:41 am, Jason <ja...@njkfrudils.plus.com> wrote: > > On Tuesday 07 June 2011 09:12:06 Cactus wrote: > > > Case has kindly tracked down another WIndows x64 bug in MPIR. > > > > > > A number of functions that return mp_bitcnt_t types were returning > > > ULONG_MAX on error rather than __GMP_BITCNT_MAX. > > > > > > SInce unsigned longs are 32-bits on WIndows x64, these ULONG_MAX > > > values were not being seen as error returns. > > > > Did you add the bitcnt type in mpir-2.3 series ? because it would make it > > and error in mpir-2.3.1 , which I suppose we should fix > > > > > I have now corrected all these erroneous return values in the MPIR > > > SVN. There is a change in gmp-impl.h but it is small and should not > > > have a serious impact on the RC1 testing that has been done for the > > > upcoming release. > > > > > > Brian > > As far as I recall, mp_bitcnt_t was in MPIR 2.3. This only impacts > on WIndows x64 so it would only be worth fixing if there are a lot of > MPIR 2.3 users depending on Windows x64.
I see how small the changes are now , it's not worth repeating the existing tests , anyway I do have a few more linux tests and win tests to do. If we are going to do a 2.3.2 we need to do it before 2.4.0 because of the sonumbers in the c++ lib (which I need to correct for trunk) The existing sonumbers for the C++ lib are not at all consistent , if you follow the rules of how to work out the sonumbers then do consider the C and C++ libs as one whole lib or as two distinct libs , looking at the existing number you can see that it alternates between the two. Jason -- You received this message because you are subscribed to the Google Groups "mpir-devel" group. To post to this group, send email to mpir-devel@googlegroups.com. To unsubscribe from this group, send email to mpir-devel+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/mpir-devel?hl=en.
