i ~never use tox, but i’d say it’s juuuust about t...
# pex
w
i ~never use tox, but i’d say it’s juuuust about time to port pantsbuild/pex to pantsbuild/pants
💯 2
h
I’m down, personally. But question, how will we handle the chicken and egg problem? If we need the Pants version in Pex to use a new patch we’re making to Pex, I don’t see how that’d be possible. We don’t want to run Pants from sources in CI Perhaps that scenario is unlikely to happen?
w
We should be able to use an older version in all cases
👍 1
f
We should be able to use an older version in all cases
Incidentally, this is how real-life chickens solve the chicken and egg problem 🐔
❤️ 2
🐔 2
🥚 1
e
The major deficiency will be caching pyenv interpreters - these are setup by tests, reused across tests and test runs and they are expensive.
👍 1
Probably requires a feature where targets can ask for something like a named cache dir.
w
yea, we should likely do that anyway. although in the interim, there is always busting out of the sandbox to store something long lived in home.