I'm seeing a repeated test failure on master for me locally in

./pants test src/python/pants/core/goals/repl_integration_test.py

, deep in some python library

checking

I saw this error message when switching dep inference to be a subprocess last week, because we were trying to load the AST as a string instead of bytes. But I fixed that and don't think it should still happen during dep inference - might be worth isolating when it fails

i don’t repro

I have not see this error on my repo.

thanks for checking - I think what might've been going on is that I forgot I still had python3.6 installed, and not 3.8 or 3.7 (but my system python is 3.9)

Makes sense, Arch just rolled 3.8 -> 3.9.

reporting a 3.9 only bug might be good

I think it might've been a 3.6 bug actually

I think that that would mean that the pants script in the pantsbuild/pants repo didn't filter out 3.6 appropriately...?

which is surprising to me - I changed it to use the same logic as the

setup/

repo that it looks for

python3.7

or

python3.8

, rather than using

python3

okay, I can actually confirm that something is broken on python 3.9 specifically

yea, that makes a bit more sense.

if I run

./pants

in the toolchain repo:

at 14:36:18 ➤ ./pants help
Using base prefix '/usr'
New python executable in /home/gregs/.cache/pants/setup/bootstrap-Linux-x86_64/pants.ThVYaB/install/bin/python3.9
Also creating executable in /home/gregs/.cache/pants/setup/bootstrap-Linux-x86_64/pants.ThVYaB/install/bin/python
Installing setuptools, pip, wheel...
done.
Collecting pip
  Using cached <https://files.pythonhosted.org/packages/ab/11/2dc62c5263d9eb322f2f028f7b56cd9d096bb8988fcf82d65fa2e4057afe/pip-20.3.1-py2.py3-none-any.whl>
Installing collected packages: pip
  Found existing installation: pip 19.0.3
    Uninstalling pip-19.0.3:
      Successfully uninstalled pip-19.0.3
Successfully installed pip-20.3.1
ERROR: Could not find a version that satisfies the requirement pantsbuild.pants==2.1.0
ERROR: No matching distribution found for pantsbuild.pants==2.1.0

ah. i don’t think we started building for 3.9 before the 2.1.x branch.

We’ve never built for Py39, and still don’t

it would be nice to provide a clearer error message