https://pantsbuild.org/ logo
Join Slack
Powered by
hey gang (<@UB2J9BQA0> in particular i think!): ou...
# announce
w
hey gang (@hundreds-father-404 in particular i think!): our validation of 
1.14.x
is trending toward completion internally
i created the 
1.14.x
milestone to start tracking the last related items: https://github.com/pantsbuild/pants/milestone/22
i don't think that resolving the go thing blocks cutting the branch
and the "`setup` in critical path" issue should land today
h
Great! Is this for this Friday?
w
i think that friday might be a good target, if the folks that care are in agreement
h
For Py3, we’re down to 3 failing targets, running Py37 in CI, and updating release.sh to close out the migration. I don’t anticipate getting that all merged by this Friday. Next Friday, maybe.
Once we have a release candidate, we only add hot fixes to it, right? Not new PRs
w
we add whatever we have consensus on, heh.
👍 1
but if one of the cherry picks to the branch is huge, it might be slightly harder to gain consensus
@hundreds-father-404: re: changing 
release.sh
... that's probably not cherry-pickable
imo
but the rest of the fixes might be
might-be/probably-are
h
Hm, so that would be a non-starter to get Py3 into this release unless we can close that by Friday, right? As we need to release a Py3 PEX to PyPi
w
*wheels
hm.
so, what is blocking starting to ship py3 wheels?
...wait. can we ship py3 wheels?
h
(Right, I keep messing that up with Pex vs wheels!) When I tried two weeks ago, we could not. We can test that out by the Build Wheels CI jobs right?
w
^
so the change would be something like... build with py2, but claim to be universal?
but/and
h
No. We would release a Py2 wheel and a Py3 wheel. Not universal.
w
why?
h
Because of that native engine issue
w
(if the 2-built wheel works under 3)
but yea, ok. either way works i suppose.
h
It does work with Py2, but last time I talked to @average-vr-56795 he thought we should not exploit that quirk that Py2 works with both. We considered always building native_engine.so with Py2 and decided it’s better to fully use Py2 and fully use Py3, not mix the two. I imagine that reasoning would extend to release.
w
i feel that that is something that we could do as part of this release, with some caveats attached
er, "before this release"
ie, we can start releasing the wheels before all errors are fixed.
h
Either way, we have to modify release script before this Friday, right?
w
yea
h
w
or monday/tuesday/etc. if what we're talking about is waiting on one more patch.
👍 1
hm... 6450 isn't quite what you were saying though.
6450 would be universal, wouldn't it?
h
Yes, agreed. The proposed changes would not release two separate wheels so we would need to tweak it Was posting issue for context with anyone following along
w
so, my feeling on the releasing of py3 wheels is that rushing to do it right before a stable release is unlikely to lead to enough time to fix everything before that stable release is completed
so, i sortof feel like cutting the stable release and then spending most of the next dev cycle recommending that people try out the py3 wheels would be better
OR, if we can get the additional wheels included in the release process sooner, that's fine too
🎉 1
there is some flexibility as to when we cut the branch, but it would be nice not to push too far past the end of the week
2 Views
Open in Slack
PreviousNext
Powered by