Dan Drake wrote: > On Tue, 01 Dec 2009 at 09:51AM +0000, Dr. David Kirkby wrote: >> My point was, since MPFR seems to be pretty good at finding compiler >> bugs, it was one worth running every time, whereas other tests may be >> less useful. > > Checking the spkg-install file, I think they *are* run every time: at > the bottom of spkg-install, I see > > # Do not bypass the checks, as some MPFR failures > # have been observed, so MPFR should be carefully tested. > cd ..; ./spkg-check > > The spkg-check file is standard, so I think those tests are getting run > on every build. > > Dan
OK. That means gcc 4.4.0 will not build Sage then, as there is a bug known in gcc 4.4.0 which will cause an MPFR test failure. That bug does not exist in earlier versions of gcc, and is fixed in later versions. (It was fixed in the gcc sources only 3 days after the release). Therefore, it would seem sensible to me that we refuse to start to build Sage with gcc 4.4.0, as we do with gcc 4.0.0. It seems silly to start to build Sage with a version of gcc, which we will know has no hope of ever building Sage. (At least without manual intervention). To my knowledge, gcc 4.4.0 is not a popular version. It is not shipped with OS X or Solaris. I can't speak for any particular linux distribution though. gcc 4.4.0 was released on 04/21/09. The bug was fixed only 3 days after the gcc release on 04/21/09 in the sources, but gcc 4.4.1 was not released until 07/22/09, so there is a 3-month period where gcc 4.4.0 version was the latest. The 'prereq' script I recently updated will exit with gcc 4.0.0. Extending that to exit with 4.4.0 would be trivial. Dave -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel-unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
