While trying to troubleshoot a test, I found that ...
# general
j
While trying to troubleshoot a test, I found that I could no longer use
repl
on our 3rdparty libraries:
Copy code
% ./pants --no-pantsd repl 3rdparty/python:zipp                                                                                                           cb/chartbeat (mybranch ⚡) beathound
Python 2.7.17 (default, Aug 19 2020, 23:21:01)
Type "copyright", "credits" or "license" for more information.

IPython 5.10.0 -- An enhanced Interactive Python.
?         -> Introduction and overview of IPython's features.
%quickref -> Quick reference.
help      -> Python's own help system.
object?   -> Details about 'object', use 'object??' for extra details.

In [1]: import zipp
---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
<ipython-input-1-eb867d7051d0> in <module>()
----> 1 import zipp

ImportError: No module named zipp

In [2]:
This works on our main branch using 1.27. Did something change with how
pants
works with
python_requirement_library
target types? I am still upgrading our
pants.toml
file so it might be something there.
w
that is unexpected. even more unexpected is that it doesn’t work on 2.0.x either
i’ll try it out on 1.30.x
j
you were able to reproduce it? I am running on 1.30.1
w
@jolly-midnight-72759: and is this on v1 or v2?
j
v1 and v2
w
with the same error?
j
doh. just worked on v1.
Ok. Ran it a few times to make sure.
./v1 --no-pantsd repl 3rdparty/python:click
works
./v2 --no-pantsd repl 3rdparty/python:click
does not work.
w
even more curious, i’m not seeing an issue in 2.0.x
j
./v2
has
./pants --no-v1 --v2 "$@"
in it
can you reproduce it with 1.30.1?
w
@jolly-midnight-72759: would you be able to file a tiny sketch of a repro on a ticket?
will try that next.
j
yup.
might be my
pants.toml
.
w
i’m also not using
ipython
, although that might not be relevant.
aha! i repro in 1.30.x with v2.
ok. so unfortunately this looks like something that was fixed between 1.30.x and 2.0.0x
j
My end goal is to get to 2.0.x. So maybe I should just hack on that version.
Is it fixed in 1.30.2x?
w
@jolly-midnight-72759: possibly… a middle ground is to get as stable as possible on each release, without releasing them.
@jolly-midnight-72759: it does not look like it, no
j
Do you still need the bug report? Sounds like it is a known issue.
w
i can file it. but would be interested in discussing the priority for you… if you are able to stably move to 2.0.x, we might not backport the fix, since we’re busying preparing for the 2.0 launch
on the other hand, if it’s going to be challenging to get on 2.0.x, we can bump that one up to help
j
I'll switch over to 2.0.x and see what I encounter there. I already have my test suite using v2.
w