What do we think for timeline of 2.9.0 and the mul...
# development
h
What do we think for timeline of 2.9.0 and the multiple user lockfile project? 🧵
We've been promising the feature since 2.7, but punted on it in 2.8 because of Go. I'm picking it up now as my main project (and lots has been done on the JVM side, yay!)
In 2.8, we blocked the release on finishing all the target modeling changes, and we realized it was probably a mistake because it was so big. Meaning harder for folks to upgrade
I think there are four main streams to do: 1) unifying everything into
generate-lockfiles
2) hooking up
compatible_resolves
3) switching to pex for lockfile generation 4) Stu's proposal for disambiguating dep inference
I imagine it will take at least 3 weeks for that all, especially the disambiguation proposal. We've already been a month since cutting dev0 Should we aim to include multiple user lockfiles in 2.9, or we're fine slipping to 2.10?
w
imo, let’s be fine slipping to 2.10.
because the JVM will not be ready to call complete in 2.9 either, and i think it’s ok to have an uneventful release in the interest of having it out on time
👍 3
h
K. Given that we're sort of using JVM as a test run for Python, I think we can prioritize step 4 of disambiguation. Without it, multiple user lockfiles aren't really feasible to use. That way, we can get feedback on how it works w/ JVM Then switch to those other tasks
w
sounds great: thank you!
i added one more comment to the doc attached to 13621, but am done touching it now
h
We should probably cut 2.9.0rc0 this Friday? It looks like we didn't start the (probably bad) practice of going into dev4 and dev5 until Pants 2.7 and 2.8 because we (I) were sprinting on the lockfile project and target changes We're supposed to be cutting rc0 every month, meaning roughly after dev3
w
irony of ironies, but: let’s wait until next week (or earlier if M2 for JVM is mostly done before then)? only 3 or 4 blockers there
i’m fairly confident it won’t slip further than that
i’m also fine not waiting, but this is … very close
h
Okay, I'm fine with waiting. I do want to avoid the situation of 2.8 having so many user API changes (mea culpa). But sounds like JVM stuff is mostly new features, and 2.9 thus far doesn't have a lot in it that's disruptive
w
definitely by end of next week, regardless of what has landed
👍 1
c
Would like to land https://github.com/pantsbuild/pants/pull/13723 as well 🙂
👍 1