<@U05GVESS708> has this spiffy PR that will make a...
# development
c
@square-psychiatrist-19087 has this spiffy PR that will make all maintainers lives easier, and has been hanging out since March. I think overall at this point it makes the most sense to land and iterate, instead of stalling the PR. But @square-psychiatrist-19087 did the right thing and wrote tests and has thus stepped on a
conftest.py
rake that causes the tests to fail in ci. I've proposed on possible fix in the PR. But if anyone has opinions on how to handle conftest.py + resolves in this case please take a look. https://github.com/pantsbuild/pants/pull/22064
h
Trying to grok the issue
c
We set
--noskip
for pytest args in
pants.ci.toml
(for everthing), but that isn't a normal pytest thing, it's something Pants created (in the root
conftest.py
file) and is only part of
python-default
. So any python tests not in
python-default
fail with 'hey what is this
--noskip
arg
h
Yep, was just playing with it
So we won’t infer a dep on conftest.py because it’s in a different resolve
So yes, I think this is the right solution
and even if it’s not perfect, I don’t want the underlying change held up any longer
But I think it’s right
It’s a little clunky, but then so is that custom resolve in the first place (do we really need it?)
c
hmm, I suppose it in part depends on how nice the
tqdm
dep is (it certainly looks nice) and then on how many new deps would accrue as the tool upgrade script expanded.
s
Thanks, I've applied the patch
👀 1