On Tue, Jul 7, 2009 at 4:57 AM, Minh Nguyen<nguyenmi...@gmail.com> wrote: > > Hi folks, > > In IRC, there was a report about the "SSSE3" and PNI flags for the > following Ubuntu binary: > > sage-4.0.2-linux-Ubuntu_9.04-i686-Linux.tar.gz > > Here's the IRC log: > > 15:39 < nifgraup> Hi,I would like to report a packaging/build bug in sage > 4.0.2, can someone guide me? > 15:45 < nifgraup> I'm running ubuntu 9.04 (kubuntu) and the > sage-4.0.2-linux-Ubuntu_9.04-i686-Linux package > 15:46 < nifgraup> and sage complains: > 15:46 < nifgraup> The following processor flags were on the build machine but > are not on this computer: > 15:46 < nifgraup> pni ssse3 > 15:47 < nifgraup> the sage-4.0.2-linux-Ubuntu_9.04-i686-Linux package should > not be built with these flags > > We've had problems with the SSE3 flag before.
Unfortunately, there has never ever been any Sage binary that was successfully built without pni + ssse3 flags and that was built on sage.math (the compile farm that builds all the sage binaries). At no point were such binaries ever released. I thought that the binaries that were explicitly labeled "sse2" that we released for sage-3.4.2 were so built but that wasn't the case as it turned out with testing. Fixing this is on the todo list, and I hope it happens for sage-4.1.1. Mostly atlas and mpir are the only two packages that "cause trouble", though I'm not convinced of that -- I tried building a binary on sage.math (on a 32-bit vmware machine), then putting it on an *old* Pentium withouts ssse3 pni, and it of course wouldn't start. Then I rebuilt atlas and mpir, and it seemed to work fine. I then ran the test suite, which did *not* come close to passing -- so other things evidently fail. If anybody wants to help with this, the plan is the following. 1. Fix the atlas and mpir spkg's so they definitely support the SAGE_FAT_BINARY environment variable (this might be the case already). 2. Build sage binary using the SAGE_FAT_BINARY variable set on a modern 32-bit linux box. 3. Move that binary to an old 32-bit linux box. 4. Try it out -- if it fails to start then the SAGE_FAT_BINARY flag doesn't work. Go to 1. 5. If Sage starts, try "random_matrix(ZZ,300).det()", which will test ATLAS. If this fails, go to 1. 6. Run the full Sage test suite. Inspect and see what is still built using ssse3/pni, etc. Go to 1. The relevant trac ticket is here: http://sagetrac.org/sage_trac/ticket/6246 -- William --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-support@googlegroups.com To unsubscribe from this group, send email to sage-support-unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-support URLs: http://www.sagemath.org -~----------~----~----~----~------~----~------~--~---
