> > > > Something must have gotten corrupted. Is there some file I can delete > > > > to start over? There isn't even an spkg/installed directory yet... > > > > Hmmm, take a look at (or re-extract) spkg/install (the main build > > > script), which sets up all package-related variables to be used in the > > > "Makefile" spkg/standard/deps, and then runs 'make'. > > > It looks fine... > > > > Or do you have different 'sed's in your PATH / perhaps in the current > > > $SAGE_ROOT/local/bin/? > > > I find that VERY unlikely, and cygcheck confirms there is just one. > > > > If your Mercurial works, you could check what changed in / happened to > > > your root repo, which includes 'install', 'deps', 'newest_version' > > > etc.; > > > > $ hg up > > > That is a problem, for reasons discussed elsewhere on sage-devel, I > > think. My working ones are all still in the 4.7.1 series: > > > abort: requirement 'dotencode' not supported! > > > But Cygwin has one already installed, luckily, which does work. hg > > diff reveals nothing (and it does work, I tested). > > > I'm going to unpack the tarball again and see what happens. > > So apparently about half of the env variables set at the beginning > have this problem that they are set to "Error running sed...". Ones > that this doesn't happen to are okay, the others are not. It really > screws up the dependencies, though, so that even ones that look okay > can fail. > > But newest_version seems to be working fine.
As you mention, it's really a bug (in newest_version) that in spkg/ install we have newest="$PKGDIR/newest_version" ATLAS=`$newest atlas` export ATLAS and we just rely on the output (a bunch of echos) being in the right format. What surprises me is that the exit 1 in newest_version is apparently not doing anything? Or at least not quitting the build. Also that the Conway polys. spkg is the only one in spkg/install that has a check for determining the package name... Any other ideas on how I can test whether "sed" is behaving properly? Again, when run by itself, newest_version behaves as expected - no failures observed in 10 or 20 tries, while I'm getting something like half the env variables for spkgs corrupted by "Error..." -- 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
