On 1/3/24 22:22, Julian Gilbey wrote:
On Fri, Dec 22, 2023 at 12:04:48AM +0100, Gregor Riepl wrote:
Hi Carsten,
[...]
My point was that if, for whatever reason, werkzeug 3.0.1 is not yet fit for
release, it should be enough to upgrade to 2.3.5 to address these unit test
failures.
[...]
That doesn't make any sense to me. These deprecations are obviously in
werkzeug and not flask-login. Why would changes in flask-login fix them?
Not saying flask-login shouldn't be updated, but that's not the right scope
for this bug report.
Just to throw in: these werkzeug failures are also causing a similar
FTBFS in debugpy; I've temporarily addressed it by skipping these unit
tests, but that's not a great solution.
I just took a quick look at upgrading the unstable/testing version of
werkzeug to 2.3.8 in the meantime. It was pretty quick, and the
package tests all pass. Shall I upload it to unstable? On salsa, I
would push to debian/2.x for the Debian branch and to upstream-2.x for
the upstream, so as not to prevent merging the experimental branch
into debian/master later. Actually, it is probably OK to use the
upstream branch for this update without causing problems, but might be
better to be safe.
(Incidentally, while doing this, I spotted one bug with the current
experimental version: it is missing a Build-Depends on
python3-markupsafe.)
Best wishes,
Julian
Hi Julian,
I do not agree with Casten here. Please do upload 2.3.8 to unstable if
it fixes things, so we have time to address all the Python 3.12 issues
right now. Please hold on uploading 3.x (as well as werkzeug) if this
can wait, because we need to address all the Python 3.12 issues first,
and now is not a good time to transition any other components.
Upgrading flask+werkzeug is *NOT* urgent. Fixing all 3.12 issues must be
our current priority.
Cheers,
Thomas Goirand (zigo)
