DevHeads.net

Postings by Matthias Klose

Focal Fossa is now open for development

Focal Fossa is now open for development, with the syncs from unstable
done and built, and autopkg testers trying to catch up.

preparing for the next archive opening

We usually tend to open the archive with a few prepared changes in place, for
the 20.04 cycle we are planning to open with

- python3.8 as a supported version, maybe already as the default.

- link time optimization enabled by default in GCC. This will
add a few ftbfs (~400). I am currently planning to have a
blacklist package listing all the source which cannot be
built with LTO.

Second Eoan Ermine test rebuilds

The second test rebuild of Eoan Ermine was started on September 06 2019 for
all architectures, all components. The rebuild of the main component is
finished, the other components (restricted, universe, multiverse) are still
building.

Unfortunately we see 1300+ build failures, and still counting ...

Python2 removal: packages which are only in Ubuntu

The removal of Python2 (build-)dependent packages, or the conversion of packages
has now been started in Debian. A few hundred are done, a few thousand are
still on the list :-/

Here is the list of source packages which are in Ubuntu only.

GCC 9 now the default in eoan

Hi,

as previously announced in

<a href="https://lists.ubuntu.com/archives/ubuntu-devel-announce/2019-June/001262.html" title="https://lists.ubuntu.com/archives/ubuntu-devel-announce/2019-June/001262.html">https://lists.ubuntu.com/archives/ubuntu-devel-announce/2019-June/001262...</a>

GCC 9 is now the default in eoan. Using the Sunday to not disturb development
during working hours with needed promotions and demotions.

Matthias

First Eoan Ermine test rebuilds

The first test rebuild of Eoan Ermine was started on June 16 2019 for
all architectures, all components.

Results (please also look at the superseded builds) can be found at

<a href="https://people.canonical.com/~doko/ftbfs-report/test-rebuild-20190614-eoan.html" title="https://people.canonical.com/~doko/ftbfs-report/test-rebuild-20190614-eoan.html">https://people.canonical.com/~doko/ftbfs-report/test-rebuild-20190614-eo...</a>

The report uses some additional color coding, marking packages different which
always failed to build, or where the build failure is no regression compared to
bionic.

Additional build failures for packages in eoan-proposed (not yet in eoan)
can be found at <a href="http://qa.ubuntuwire.com/ftbfs/" title="http://qa.ubuntuwire.com/ftbfs/">http://qa.ubuntuwire.com/ftbfs/</a>

Please help fixing the build failures.

There is also

Delegating hinting for autopkg failures to core developers and MOTU

Please can we delegate the hinting for autopkg test failures to core developers
and MOTU?

When ignoring an autopkg test failure, you usually have a reason to do so. As a
core developer you already can work around autopkg test failures with uploading
a work-around in a new package version, both for main and universe packages.
The disadvantage is a longer turn-around, re-triggered autopkg tests,
introducing a delta which has to be maintained.

Cosmic test rebuild

Here are results for a cosmic test rebuild (main only), combining the release
and the updates pocket:

<a href="http://people.canonical.com/~doko/ftbfs-report/test-rebuild-20190404-cosmic.html" title="http://people.canonical.com/~doko/ftbfs-report/test-rebuild-20190404-cosmic.html">http://people.canonical.com/~doko/ftbfs-report/test-rebuild-20190404-cos...</a>

Bug reports are filed with tags ftbfs and rls-cc-incoming. No new reports are
filed where the issue still exists in the most recent disco test rebuild.

Matthias

Second Disco Dingo test rebuild

The second test rebuild of Disco Dingo was started on April 04 2019 for
all architectures, just the main component.

Disco Dingo test rebuilds

The first test rebuild of Disco Dingo was started on December 21 2018 for
all architectures, all components.

Cosmic Cuttlefish test rebuilds

The first test rebuild of Cosmic Cuttlefish was started on September 11 2018 for
all architectures, all components (main component and seeded packages finished,
unseeded packages still building).

xenial (16.04 LTS) and bionic (18.04 LTS) test rebuilds

There are four test rebuilds done for xenial and bionic (main only), both the
release and the updates pockets. The results of both builds are combined in one
table. Please see

<a href="http://people.canonical.com/~doko/ftbfs-report/test-rebuild-20180727-main-xenial.html" title="http://people.canonical.com/~doko/ftbfs-report/test-rebuild-20180727-main-xenial.html">http://people.canonical.com/~doko/ftbfs-report/test-rebuild-20180727-mai...</a>

<a href="http://people.canonical.com/~doko/ftbfs-report/test-rebuild-20180730-bionic.html" title="http://people.canonical.com/~doko/ftbfs-report/test-rebuild-20180730-bionic.html">http://people.canonical.com/~doko/ftbfs-report/test-rebuild-20180730-bio...</a>

The table also has some new color codes to mark packages which always failed to
build, and for packages which failed to build in a reference release.

Cosmic Cuttlefish is now open for development

Cosmic CANIMAL [1] is now open for development, with the syncs from unstable
done and built. The development version starts with only a few changes:

- The final GCC 8.1.0 release (not yet used as the default)

- Only one transition almost done (ncurses)

Please be aware about the planned compiler changes (GCC update from 7 to 8)
around June/July, an update to OpenJDK 11, and an update from Python 3.6 to 3.7.

Please check your uploads in a cosmic chroot, don't just test in a bionic or
artful environment. See [2] or [3] how to setup such a development chroot.

Second Bionic Beaver test rebuilds

The second test rebuild of Bionic Beaver was started on April 08 2018 for all
architectures, all components (main component and seeded packages finished,
unseeded packages still building).

1000 New Year's pledges for 2018 (first Bionic Beaver test rebuilds)

The first test rebuild of Bionic Beaver was started on December 21 2017 for all
architectures, all components (main component and seeded packages finished,
unseeded packages still building). The number of build time failures
unfortunately is at an all time high around a four digit number.

Bionic Beaver now open for development

Before heading on with Bionic Beaver, let's have a look at all the remaining
build failures in Artful Aardvark [1], which is the highest number of build
failures at the end of a release cycle we ever had. Please don't repeat that
and plan to address these build failures in time for the Bionic Beaver release
(at least for all the packages in main and in seeds).

Bionic Beaver [2] is now open for development.

Second Ubuntu Artful Aardvark test rebuild

The second test rebuild of Artful Aardvark was started on September 22 for all
architectures, all components (main component and seeded packages finished,
unseeded packages still building).

For arm64 and armhf the gcc-6 and gcc-7 packages are based on the Linaro
6-2017.09 snapshots.

Results (please also look at the superseded builds) can be found at

<a href="http://qa.ubuntuwire.org/ftbfs/rebuilds/test-rebuild-20170922-artful.html" title="http://qa.ubuntuwire.org/ftbfs/rebuilds/test-rebuild-20170922-artful.html">http://qa.ubuntuwire.org/ftbfs/rebuilds/test-rebuild-20170922-artful.html</a>

The test rebuild uses the release pocket only, without packages from the
proposed pocket.

Additional build failures for packages in artful-proposed (not yet in artful)

GCC 7 default update planned for early August

Hi,

it's this time of the year where you go to your vacation, and look at a new
shiny new GCC version in the archive when you come back. Well, not exactly ...

I'm planning to change the default on Aug 03/04, and then working to get things
migrated and buildable again, including two transitions for GCC runtime
libraries. Fixing things pro-actively is appreciated. If there are questions,
please use the #ubuntu-devel irc channel.

First Ubuntu Artful Aardvark test rebuilds (and upcoming GCC 7 transition)

The first test rebuild of Artful Aardvark was started on July 6 for all
architectures, all components (main component and seeded packages finished,
unseeded packages still building).

For arm64 and armhf the gcc-6 packages are based on the Linaro 6-2017.06
snapshot.

Results (please also look at the superseded builds) can be found at

<a href="http://qa.ubuntuwire.org/ftbfs/rebuilds/test-rebuild-20170706-artful.html" title="http://qa.ubuntuwire.org/ftbfs/rebuilds/test-rebuild-20170706-artful.html">http://qa.ubuntuwire.org/ftbfs/rebuilds/test-rebuild-20170706-artful.html</a>

The test rebuild uses binutils, gcc-6, gcc-7 and linux packages currently still
in the proposed release pocket, which can also be found separately in the PPA

<a href="https://launchpad.net/~ub" title="https://launchpad.net/~ub">https://launchpad.net/~ub</a>

Second Ubuntu Zesty Zapus test rebuilds (all components, all architectures, GCC 7)

The second test rebuild of Zesty Zapus was started on March 22 for all
architectures, all components (main component and seeded packages finished,
unseeded packages still building).

For arm64 and armhf the gcc-6 packages are based on the Linaro 6-2017.03
snapshot.

Results (please also look at the superseded builds) can be found at

<a href="http://qa.ubuntuwire.org/ftbfs/rebuilds/test-rebuild-20170322.1-zesty.html" title="http://qa.ubuntuwire.org/ftbfs/rebuilds/test-rebuild-20170322.1-zesty.html">http://qa.ubuntuwire.org/ftbfs/rebuilds/test-rebuild-20170322.1-zesty.html</a>

Additional build failures for packages in zesty-proposed (not yet in zesty)
can be found at <a href="http://qa.ubuntuwire.com/ftbfs/" title="http://qa.ubuntuwire.com/ftbfs/">http://qa.ubuntuwire.com/ftbfs/</a>

Please help fixing the build failures.

Another test rebuild u

First Ubuntu Zesty Zapus test rebuilds (all components, all architectures, GCC 7)

The first test rebuild of Zesty Zapus was started on December 02 for all
architectures, all components (main component and seeded packages finished,
unseeded packages still building).

For arm64 and armhf the gcc-6 packages are based on the Linaro 6-2016.10
snapshot.

Results (please also look at the superseded builds) can be found at

<a href="http://qa.ubuntuwire.org/ftbfs/rebuilds/test-rebuild-20161202-zesty.html" title="http://qa.ubuntuwire.org/ftbfs/rebuilds/test-rebuild-20161202-zesty.html">http://qa.ubuntuwire.org/ftbfs/rebuilds/test-rebuild-20161202-zesty.html</a>

Additional build failures for packages in zesty-proposed (not yet in zesty)
can be found at <a href="http://qa.ubuntuwire.com/ftbfs/" title="http://qa.ubuntuwire.com/ftbfs/">http://qa.ubuntuwire.com/ftbfs/</a>

Please help fixing the build failures.

Another test rebuild u

Zesty Zapus open for development

Zesty Zapus [1] is now open for development, with syncs from unstable
still running. The development version already starts with some
changes:

- GCC on armhf and arm64 now is built again from the GCC Linaro branch.

- Starting with library transitions: openmpi, boost 1.62.

Please check your uploads in a zesty chroot, don't just test in a xenial or
yakkety environment. See [2] or [3] how to setup such a development chroot.

Second Ubuntu Yakkety Yak test rebuilds (all components, all architectures, Linaro-GCC)

The second test rebuild of Yakkety Yak was started on September 16 for all
architectures, all components (main component and seeded packages mostly
finished, unseeded packages still building).

Results (please also look at the superseded builds) can be found at

<a href="http://qa.ubuntuwire.org/ftbfs/test-rebuild-20160916-yakkety.html" title="http://qa.ubuntuwire.org/ftbfs/test-rebuild-20160916-yakkety.html">http://qa.ubuntuwire.org/ftbfs/test-rebuild-20160916-yakkety.html</a>

Additional build failures for packages in yakkety-proposed (not yet in yakkety)
can be found at <a href="http://qa.ubuntuwire.com/ftbfs/" title="http://qa.ubuntuwire.com/ftbfs/">http://qa.ubuntuwire.com/ftbfs/</a>

Please help fixing the build failures.

Another test rebuild using the Linaro-GCC 6 as the default (planning the
defaults change around th

First Ubuntu Yakkety Yak test rebuilds (all components, all architectures, GCC 6)

The first test rebuild of Yakkety Yak was started on July 01 for all
architectures, all components (main component and seeded packages finished,
unseeded packages still building).

Results (please also look at the superseded builds) can be found at

<a href="http://people.ubuntuwire.org/~wgrant/rebuild-ftbfs-test/test-rebuild-20160701-yakkety.html" title="http://people.ubuntuwire.org/~wgrant/rebuild-ftbfs-test/test-rebuild-20160701-yakkety.html">http://people.ubuntuwire.org/~wgrant/rebuild-ftbfs-test/test-rebuild-201...</a>

Another test rebuild using GCC 6 as the default (planning the defaults change
around the end of July) can be found at

<a href="http://people.ubuntuwire.org/~wgrant/rebuild-ftbfs-test/test-rebuild-20160705-gcc6-yakkety.html" title="http://people.ubuntuwire.org/~wgrant/rebuild-ftbfs-test/test-rebuild-20160705-gcc6-yakkety.html">http://people.ubuntuwire.org/~wgrant/rebuild-ftbfs-test/test-rebuild-201...</a>

Please start fixing build failures with GC

GCC/binutils and Python 2/3 updates for xenial

In the past weeks I've prepared updates for 16.04 LTS, updating gcc-5, binutils,
python2.7 and python3.5 to the last upstream point releases. The archive was
rebuilt using these versions, and an analysis of the test rebuild can be found
at LP: #1586673 (SRU for binutils and gcc-5).

binutils is already available in xenial-proposed, the other packages can be
found in the PPA ubuntu-toolchain-r/ppa.

OpenGL to OpenGL ES change on arm64?

built with OpenGL ES. Do we really have to cripple an architecture like this?
I don't see any discussion about this. How does this affect things like GPU
accelerators and CUDA aware packages?

Matthias

Yakkety Yak is now open for development

Yakkety Yak [1] is now open for development, with syncs from unstable
still running. The development version already starts with some
changes:

- GCC is configured to build with -fPIE (and to pass -z now to ld)
by default on amd64 and ppc64el.

- Three library transitions are almost done: boost 1.60, ICU 57,
and libpng 1.6.

- A bunch of library transitions is already triggered by the sync
from Debian.

oracle-java8-installer again inflates the priorities for the alternatives

Hi,

The oracle-java8-installer package in the webupd8team PPA again inflates the
priorities for the alternatives, as in looking for the highest available
alternative, then adding one to it and using that as the new alternative value.
This is not how it should be, and even wrong when you install multiple
oracle-java*-installer packages. You seem to have an oracle-java8-set-default
package which explicitly should do that, don't mess around with the priorities
for the alternatives. Pretty please fix this in a way that you don't just
inflate the priority of the alternatives.

Third Ubuntu Xenial Xerus test rebuilds (all components, all architectures)

The third test rebuild of Xenial Xerus was started on April 01 (no joke) for all
architectures, all components (it is finished besides some pending builds for
powerpc and armhf). Compared to the last test rebuild this sees some new build
failures introduced by new upstream versions during the feature freeze (e.g.

Second Ubuntu Xenial Xerus test rebuilds (all components, all architectures)

The second test rebuild of Xenial Xerus was started in late February for all
architectures, all components (it is finished besides some pending builds for
powerpc and armhf).

Results (please also look at the superseded builds) can be found at

<a href="http://people.ubuntuwire.org/~wgrant/rebuild-ftbfs-test/test-rebuild-20160226-xenial.html" title="http://people.ubuntuwire.org/~wgrant/rebuild-ftbfs-test/test-rebuild-20160226-xenial.html">http://people.ubuntuwire.org/~wgrant/rebuild-ftbfs-test/test-rebuild-201...</a>

For arm64 and armhf the gcc-5 packages are based on the Linaro 5-2016.02
snapshot.

Additional build failures for packages in xenial-proposed (not yet in xenial)
can be found at <a href="http://qa.ubuntuwire.com/ftbfs/" title="http://qa.ubuntuwire.com/ftbfs/">http://qa.ubuntuwire.com/ftbfs/</a>

Please help fixing the build failures.

Matthias