Version 1.12.53-1 added a --with-long-double option to configure, but
this FTBFS on ppc64el due to (I guess) some differences in how long
doubles are defined on that architecture, which breaks some macros that
use the GNM_const() macro. I've filed update excuse bug #1991706.
It wasn't obvious to me how the macro itself should be fixed, but for
now I've just reverted the newly added --with-long-double option. I
notice Debian has the same build failure so we can re-enable the option
when this gets further investigation.
At time of writing, gnumeric was still in the unapproved queue
(https://launchpad.net/ubuntu/kinetic/+queue?queue_state=1&queue_text=gnumeric)
however I've verified in a PPA that it builds ok now
(https://launchpad.net/~bryce/+archive/ubuntu/gnumeric-fix-lp1991706)
so I believe gnumeric 1.12.53-1ubuntu1 should migrate successfully.
### node-resolve (and node-*) ###
There are a handful of node-* packages that seem interdependent with
each other. I retriggered the lot of them, and at least that got
node-resolve to successfully migrate. Seems to have partially cleared a
few others.
node-qs and node-yarnpkg test failures are blocking node-deep-equal, and
once that migrates it looks like a lot of other node-* packages should
migrate as well. Their test logs show they're blocked on missing
modules but I think that may now be resolved. I've retriggered these
three packages in combination together in case they're mutually
dependent (the test logs indicate dependency issues rather than actual
test code failure).
### Loggerhead ###
2.0.0-1 was supposed to include a fix for the autopkgtest issue we saw
regarding missing modules, however we still see the same issue in our
testing. Debian released 2.0.0-2 which I gather fixed an issue
preventing the tests from running. I've syncpackage'd this in for us,
however from Debian's CI it looks like it may still get hung up on
pygments; if that also happens for us, this may need to have
python3-pygments added as a Depends in d/t/control for the autopkgtest.
I filed update-excuse bug LP: #1991613 with the above info.
### scikit-learn ###
This package hits a 'bus error' on armhf. This issue seems not to have
a bug report in Debian, however bus errors are mentioned on both Deb:
#1008369 and #1003165. Both bugs have extensive discussion, however it's
unclear if a fix is in sight; main suggestions appear to be to drop the
architecture or skip the test case. Those may not be this same issue
though; it seems "Bus error" is a generic error that's been happening
for other tests on armhf. Upstream is also aware of armhf tests being
in a really bad state.
A previous +1'r also suggested skipping the failing test, so I've gone
ahead and added test_dist_metrics to appropriate sections of both
d/rules and d/t/python3, and uploaded this as 1.1.2+dfsg-5ubuntu1.
Since the testsuite bails as soon as it hits the first bus error, it's
possible there will be subsequent tests failing the same way, in which
case maybe just keep adding excludes.
I've filed update-excuse bug LP: #1991621 with the above info.
### update-manager / gnome-flashback ###
Retriggered and it passed and migrated.
https://autopkgtest.ubuntu.com/packages/update-manager/kinetic/armhf
### lastz ###
FTBFS on riscv64. A simple rebuild succeeded, and enabled the package
to migrate successfully.
### snowflake ###
FTBFS on riscv64. A simple rebuild succeeded, and enabled the package
to migrate successfully.
### vectorscan ###
Filed update excuse LP: #1991622 with notes from previous +1'rs.
Debian sees the same issue and already has a bug report.
### gensio / ser2net ###
I filed update-excuse LP: #1991617 with the notes from previous +1'rs
but didn't have time to dig into it.
Bryce
--
ubuntu-devel mailing list
ubuntu-devel@lists.ubuntu.com
Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel
