Hi <@U051221NF>, I'm curious if there are plans to...
# general
e
Hi @happy-kitchen-89482, I'm curious if there are plans to tackle https://github.com/pantsbuild/pants/issues/15704. It's a pretty big pain point for us right now, that adding a dep will cause many unrelated deps to be upgraded.
h
I am implementing this in the new python backend. It could be implemented in the existing backend, if someone is motivated to take it on. As was mentioned in that ticket, Pex already supports it.
e
ah, cool, thanks for that link. Of course my next question is "when is that backend going to land" which I realize is rather impossible to answer 😬 . But I'm curious about how I can keep up with progress.
h
I’m actually making progress on it right now. The first thing I’m tackling is resolves, lockfiles and so on.
I’m hoping to be able to get the new backend to delegate to the old one for the mundane things like running pytest, linters, packaging etc, at least initially. The main thing is to integrate the new Pex features, and possible uv.
e
nice. Yeah I saw the
uv
discussion, that would be great too if it speeds up lockfile generation.
f
in case this behavior of pex in https://github.com/pex-tool/pex/pull/1799 can be triggered by passing a pex argument, then you may want to experiment using newly added feature that can allow passing global args to pex. https://github.com/pantsbuild/pants/pull/21202 I've added it to avoid extending Pants whenever there's a helpful flag from pex I'd like to use in a given command.
e
oh cool, that seems very helpful. I am a bit unsure of how to get Pex to do the "no update" behavior but I will poke around.
c
There are some subtle variations in the use cases in that ticket, and several reasonable things people would like to do. Is adding a new dep (as opposed to removing one, or changing the version of just one dep) your largest painpoint?
e
yes, adding a new dep is the biggest for sure. The others do come up as well though.
🙏 1