On 10/27/2010 08:26 PM, Mitesh Patel wrote: > On 10/27/2010 08:43 AM, Dr. David Kirkby wrote: >> 1) Despite previously claiming in README.txt that we supported Fedora >> 32-bit on x86, the only host with Fedora 32-bit, >> >> http://build.sagemath.org/sage/buildslaves/cicero-1 >> >> has never had a successful build and test. >> >> 2) Despite previously claiming we supported Fedora 64-bit on x86, the >> only Opteron machine >> >> http://build.sagemath.org/sage/buildslaves/flavius-1 >> >> has never had a successful build+test. Perhaps there's an Opteron >> specific bug here. > > I've already posted to the (non-public) sage-skynet mailing list about the > reproducible doctest errors on [several of] the Skynet Fedora 13 machines. > The errors all appear to stem from building on an NFS filesystem, on > which SQLite may be unable to lock the graph database graphs.db.
I've worked around this by copying graphs.db to /tmp and symlinking the copy to SAGE_ROOT/data/graphs/graphs.db. I've also used DOT_SAGE=`mktemp -d /tmp/dot_sage.XXXXXXXXXX` to avoid similar errors in database.py, whose doctests make databases under tmp_dir()s. I don't think mktemp is a POSIX command, but it seems to be widespread: http://stackoverflow.com/questions/2792675/how-portable-is-mktemp1 -- 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
