LTS & ELTS Report for December 2024
In November I spent 0 hours on Debian LTS, and 2.5 on ELTS.
I reviewed updates by Bastien Roucariès for Python 2.7, and 3.4. I helped to get the packages to build under Salsa CI.
In November I spent 0 hours on Debian LTS, and 2.5 on ELTS.
I reviewed updates by Bastien Roucariès for Python 2.7, and 3.4. I helped to get the packages to build under Salsa CI.
In November I spent 0 hours on Debian LTS, and 5.5 on ELTS.
I reviewed updates by Bastien Roucariès for Python 2.7, 3.5, and 3.7. This mostly involved backporting the same set of patches to older releases, where things worked a little differently.
In October I spent 1 hours on Debian LTS, and 2.5 on ELTS.
Updating the distro-info-data
database. A straightforward data update,
but I ran into a bug in sbuild
+autopkgtest
that made my sbuild
unable to build on old ELTS releases. Once understood, it was trivial
to fix.
I also re-investigated the current state of the PEP-440 issue that affects Python virtulenvs on older releases, that have system site package exposed to the virtualenv. Newer Python tooling (likely to be installed in a virtualenv) does not accept that Python packages can have versions that don't parse under PEP-440's rules. Some Debian packages include nonsense Python package versions, because it didn't matter at the time. For supported packages, we have corrected this.
In June I spent 0.5 hours on Debian LTS, and none on ELTS.
Updating the distro-info-data
database, which turned out to be no
update at all.
In preparing to announce the end-of-life of Debian 10 LTS, the date
shifted a couple of times before ending back where it started.
In April I spent 1.5 hours on Debian LTS and 1.5 hours on Debian ELTS:
Updating the distro-info-data
database.
This update included the new columns that had previously been stripped
out of distro-info-data
updates to older releases.
That required groundwork in October 2023,
updating the older distro-info
versions to handle unexpected columns.
In October I spent 10 hours on Debian LTS:
Investigating the number of packages with invalid PEP-440 versions. These are problematic when using newer setuptools or pip on an older system.
Updating the distro-info-data
database.
This was more complex than usual, because the update changed a number
of historical dates, breaking the distro-info
test-suite.
Updating distro-info
to account for the date changes above, and
while we were at it:
distro-info-data
updates in the future.In October I spent 3 hours on Debian ELTS, on the same
distro-info-data
and distro-info
update tasks.
In April I spent 1 hour on Debian LTS, and 1.5 hours on Debian ELTS, for the same task:
distro-info-data
database.My Freexian-sponsored Debian work is now reported on in the Freexian blog:
In January I spent 4.5 hours on Debian LTS, resulting in no uploads:
In January I spent 0 hours on Debian ELTS. I'm cutting my LTS & ELTS hours down to 0, for the next couple of months, as I'm sailing and have more limited time.
During the month, Freexian sponsored 8 hours of 37 I spent on Debian:
A Freexian customer also sponsored some work on Debian packages:
In December I spent 2.5 hours on Debian LTS, resulting in no uploads:
In December I spent 0 hours on Debian ELTS.
During the month, Freexian sponsored 7.5 hours of 36 I spent on Debian:
--remaining-packages
,
multiple libraries in a source package, and
removing Python 2.7 / PyPy 2.7 support.
Landing Antonio Terceiro's changes to add an automatic test
runner, pybuild-autopkgtest
,
and some follow-up patches to it.
Supporting data installation in PEP517 builds.In November I spent 2.5 hours on Debian LTS, resulting in no uploads:
In November I spent 0 hours on Debian ELTS.
During the month, Freexian sponsored 16 hours of 80 hours I spent on Debian:
debhelper.log
files, in overrides, to support
--remaining-packages.
Rebased the pybuild-autopkgtest branch to prepare to land it.
Tested it on several of my packages.