Bugzilla Bug 182196

Dear arch-team-members, please re-keyword dev-utils/boost-build-1.34.0 and
dev-libs/boost-1.34.0 for your arch.

Please remember that the tests will take several hours and nearly 1GB of
disk-space and the ebuild will not fail on test-failures.
This policy is recommended by upstream.
After running the tests, there will be a file
"${ROOT}usr/share/doc/${PF}/status/cs-$(uname).html" which is a short summary
of all the tests, if most of them succeeded, you can safely keyword it.

If there are many failures, please write a comment here and we'll see how to
proceed.

boost-build is good for ~sparc.  However, boost-1.34.0 ignores user CXXFLAGS,
and on sparc this is very bad.  Hence, I'm holding ~sparc for boost until that
is explained.  In this case, I have '-mcpu=ultrasparc3' and I really mean it
unless there is a reason not to. Bug 59506 explains everything.

Hmm, that shouldn't happen.
arch-teams: please hold on.

(In reply to comment #1)
> boost-build is good for ~sparc.  However, boost-1.34.0 ignores user CXXFLAGS,
> and on sparc this is very bad.  Hence, I'm holding ~sparc for boost until that
> is explained.  In this case, I have '-mcpu=ultrasparc3' and I really mean it
> unless there is a reason not to. Bug 59506 explains everything.
> 

I notice that the ebuild forces optimization=none, and it thinks it is building
no optimization.  Looking more carefully, however, I see that the call to
gcc.compile.c++ actually ends up using my CXXFLAGS:
Most tests pass, so adding ~sparc keyword.  My original complaint was
incorrect, based on misreading what I was seeing.

~sparc is good, and sorry for the noise.

Most of the python tests fail, others hardly. Added ~ppc64.

~alpha/~ia64 done

Marked ~hppa.

This package is nuts.  The tests have been running for 13 hours already, and no
sign of ending is in sight....

I see random mentions of "FATAL"'s, "failed..." and "skipped", so I'm getting a
tad worried about this package.  Ah well.

@kumba: hehe, I told you :-) Well, some of the tests are meant to fail and mips
worked well last time I tested a CVS snapshot on swarm.

(In reply to comment #7)
> This package is nuts.  The tests have been running for 13 hours already, and no
> sign of ending is in sight....
> 
> I see random mentions of "FATAL"'s, "failed..." and "skipped", so I'm getting a
> tad worried about this package.  Ah well.
> 
On sparc SB1000 (1x900 + 1x750), tests ran for over 24 hours, several failures
in the python tests (as Markus noted in Comment #4).

I'm seeing lots of failures on ppc too, but I've marked it ~ppc as well.

I got to the end (finally -- TMPDIR ran out of space cause I was dumb and had
it on a tmpfs mount), but it seems the XML log generation flunked.

...failed updating 25 targets...
...skipped 35 targets...
...updated 4823 targets...
boost_root: /usr/obj/portage/dev-libs/boost-1.34.0/work/boost_1_34_0
locate_root: /usr/obj/portage/dev-libs/boost-1.34.0/work/boost_1_34_0
no errors detected
Using
/usr/obj/portage/dev-libs/boost-1.34.0/work/boost_1_34_0/bin.v2/status/rational_test.test
to determine compilers

**** exception(205): std::string: xml: unexpected eof
******** errors detected; see standard output for details ********

!!! ERROR: dev-libs/boost-1.34.0 failed.


All in all, it looks clean to me though.  I'll mark ~mips if you can't think of
any reason to retrieve and view the log.  And I'll hope 1.35.0 is a year or
more off :P

unstable on mips for both boost and boost-build.

ping ?