I am running `pants generate-lockfiles` and it jus...
# general
c
I am running
pants generate-lockfiles
and it just sits there counting up the time, it's at over 5 minutes now. How can I see what the problem is?
-ldebug
shows that the pex command is started, but nothing more than that....
a
What version of pants? And what’s the message it’s showing when counting up time
c
v 2.13.0.dev5 ⠙ 32.73s Generate lockfile for foo
a
In a large repo, that takes a long time, but yeah, it’s not the problem we had in 2.12
c
It's a small repo, just added a couple of dummy files. I tried with the latest 2.15 too, seems the same. I will let it run over lunch, but something feels wrong.
It looks like you were right, it just takes a loooooooooong time to finish, which I don't understand why. There should at least be a warning in the docs "even for a small project with no dependencies, it will take 10 minutes or so for unknown reasons" 😉 Thanks!
b
One underlying issue relates to the Python ecosystem. There's no way to know what dependencies a project has without downloading the entire package. It's honestly a glaring hole I hope PyPA can address one day.
h
Could you put a repro of this in a public GitHub repo? If the sources are dummy files then I assume this won't expose any proprietary stuff. I'd like to verify that this is a pathology of pip, not of pants...
👍 1
Because with no external dependencies lockfile generation should be trivial
So I'd love to see what's going on there
h
We also expect performance to improve soon because Pex recently gained the ability to use newer versions of pip, which have performance improvements for its dependency resolution algorithm. We only need to finish wiring that up to Pants
w
2.13.0
is released, and contains lots of patches (to PEX in particular) beyond what is in the
<http://2.13.0.dev|2.13.0.dev>
releases
we generally do not recommend using dev releases unless you are looking for a particular unreleased feature
c
Hi all, thanks for all the input. I used the version recommended in the docs, I thought 0.dev5 is after .0. Anyway, to the matter at hand, I have to apologize for the confusion - it looks like it was some weird networking issue. For some reason, in the terminal I was using, a http proxy was configured and it behaved poorly. I tried in a fresh terminal and it is no longer slow. So all is good!
w
I used the version recommended in the docs
hm, yikes! i’ll get a patch out to fix that. thanks. we recently switched to a new docsite strategy, and that mechanism is stale.