dev-lang/wml-2.0.11 has been in the tree for about a month and has no open bugs. This is unlike the currently stable version 2.0.9-r1 which does not build in many cases. I know that wml-2.0.11 is not multilib-strict due to weird Makefile b0rkage, but this is not something that is easily fixed. Besides, this is broken in 2.0.9 also so I hope the 64-bit arches will look the other way on this one. Good luck in building and testing this fickle beast!
It fails 19 out of 20 tests. Is that expected? Do you need more information?
For me it fails just a single test (test 3 in 11-wmk.t), and that seems to be a simple syntax issue which is easily fixed. I've run the tests by issues 'make test' in /var/tmp/portage/dev-lang/wml-2.0.11/work/wml-2.0.11. Some more details would be useful in tracking this down. I don't actually care so much about the tests because I don't think they are very reliable, but so many failures may point to a build issue that would also manifest on installation.
These failures seem to be related to the way in which FEATURES="test" is running the tests. ebuild.sh actually runs 'make check -n' first as it naively believes that this should not cause any harm. But wml's build system is weird enough that this actually does run part of the tests, and this in turn causes the actual 'make check' to fail because the freshly compiled wml binaries are no longer copied into the test hierarchy. Testers: it would be great if you could manually run the tests instead of using FEATURES="test"
(In reply to comment #3) > Testers: it would be great if you could manually run the tests instead of using > FEATURES="test" Could you redefine src_test()?
dev-lang/wml-2.0.11-r1 is now in CVS with a custom src_test. This should make testing for arch testers easier. Note that test 3 in 11-wmk might fail but this is not serious enough to hold up this release. Since this is the only change I would like to ask for stabilization of wml-2.0.11-r1 instead of wml-2.0.11.
x86/amd64 stable
Stable on ppc
ia64 stable
sparc stable.
