Hi Lucas, > > OpenMPI has a serious bug that prevents me from using it atm (it's > > fixed upstream). > > Can you give a pointer to a bug report or something? If the bug is > serious not just for you, it might be possible to backport the fix to > Debian.
I must admit that it was a while since I last checked for this bug. I just had a look again and in the current debian/testing version it should be fixed. The problem occured when you used composite types containing other composite types. A scenary that shouldn't occure too often. > > Actually I don't care which MPI-documentation I have installed. As > > this problem persists I install lam4-doc. But the recommendation of > > mpich2-doc results in the removal of lam4-dev as the prefered solution > > of apt. This behaviour is imho strange for the average user. > > Yes, since it's only a "recommend", you could probably remove mpich2-doc > and install lam4-dev back. I've changed the bug title accordingly, and > I'll push the change -- I'll wait a few days first to see if something > else shows up. I think this should do the trick. > > PS: Wouldn't a NMU of lam be a better solution to the manpage problem > > than this conflict? > > It has already been NMUed 5 times... I'd prefer to invest time to make > OpenMPI and MPICH2 rock in Debian, so we can phase out LAM/MPI and > mpich1, rather than work on LAM and mpich... That is true. Cheers Christian -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
