[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#975535: elpy's autopkg tests fail with Python 3.9

Hi Adrian,

Thank you for checking in with this bug!  Please let me know ASAP if
another autoremoval exception will be provided, because if necessary I
can do the shady thing of disabling tests to buy time...but I'd really
prefer not to!  My primary upstream contacts appear to be on holiday
and/or seem to be unavailable for some reason, but the upstream
community has stepped forward to fix these issues.  My primary worry is
that this won't happen fast enough to prevent autoremoval during the
next phase of the freeze, with no reentry into testing.

Adrian Bunk <bunk@debian.org> writes:

> On Fri, Dec 25, 2020 at 08:57:49PM -0500, Nicholas D Steeves wrote:
>> I've made progress with this, and encountered a couple blockers along
>> the way (one outstanding at this time).  Currently the need for Jedi
>> 0.18 means the work in git (Elpy 1.35 minus rope, plus cherrypicked
>> support for Jedi refactoring) cannot yet be uploaded.
> Jedi 0.18.0 is now in unstable:
> https://tracker.debian.org/pkg/python-jedi

Unfortunately it looks like the upstreams PRs for Jedi 0.18 support are
insufficient/incomplete, and I'll need more time to work on bugs
introduced by dependencies that introduced various incompatible

The state of Elpy is worse than I'd like to see...on my moderately
patched local HEAD (based on upstream master branch) I'm seeing upwards
of 27 tests that must be skipped.  It also looks like Python 3.9 may
have made breaking changes to how pdb works.

The good news is 438/465 tests are still good (on my local HEAD), so on
the whole it's better to have Elpy in Bullseye than to not :-)  Version
1.34 has 423/436 passing.  IIRC 1.35 has only seven failing tests, but
less coverage than upstream master's HEAD.

At this point I'm using an upstream snapshot in the Debian git project
for Elpy, because staying close to the next upstream release will make
cherry picking any fixes much easier, and I've recently seen a
nontrivial amount of code churn and refactoring.

My hope is that the PR discussed at the following link will provide the
solution for bullseye: https://github.com/jorgenschaefer/elpy/issues/1868


Attachment: signature.asc
Description: PGP signature

Reply to: