by

Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • 02:07
    jreback milestoned #36510
  • 02:07
    jreback labeled #36510
  • 02:07
    jreback labeled #36510
  • 02:07
    jreback labeled #36510
  • 02:07
    jreback labeled #36510
  • 02:07
    jreback labeled #36510
  • 02:03
    arw2019 synchronize #35751
  • 01:59
    jbrockmendel labeled #11164
  • 01:59
    jbrockmendel labeled #11164
  • 01:58
    jbrockmendel labeled #10539
  • 01:58
    jbrockmendel labeled #9893
  • 01:58
    jbrockmendel commented #9893
  • 01:57
    jbrockmendel labeled #10175
  • 01:53
    jbrockmendel labeled #8995
  • 01:52
    jreback commented #36538
  • 01:52
    jreback milestoned #36538
  • 01:52
    jreback labeled #36538
  • 01:52
    jreback labeled #36538
  • 01:51
    jreback commented #36537
  • 01:51
    jbrockmendel commented #7308
Daniel Saxton
@dsaxton
does anyone here have any ideas on keeping track of stale PRs? i noticed there's a GitHub action for dealing with these and i wonder if that could make sense to use in pandas: https://github.com/actions/stale (might be some cases where a PR is stale but would be preferable not to be closed, but the owner can always reopen later; also probably would want to disable any closing of old issues since there seem to be a lot of unresolved ones that should remain open even if no activity)
2 replies
Erfan Nariman
@erfannariman
When validating a single dostring I get ImportError: cannot import name 'VALID_CLOSED' from 'pandas._libs.interval', this happens in core/arrays/interval
I am running the following line: >python scripts\validate_docstrings.py pandas.DataFrame.lookup
Rebuilding did not help, anyone an idea what might solve this?
Marco Gorelli
@MarcoGorelli

Rebuilding did not help, anyone an idea what might solve this?

Works for me, all I can think to advise is git fetch upstream master, get merge upstream/master, python setup.py build_ext -i -j 8

Erfan Nariman
@erfannariman
What would be the way to build requirements-dev.txt from the conda environment.yml. I tried generate_pip_deps_from_conda.py, but this seems to do nothing.
1 reply
Erfan Nariman
@erfannariman
Can someone with the correct rights retrigger the pipelines on #36356 , the failed tests seem unrelated with the changes in the PR.
Erfan Nariman
@erfannariman
Thanks @MarcoGorelli
Erfan Nariman
@erfannariman
Can someone retrigger travis on #35244 please? Thanks
Daniel Saxton
@dsaxton
has anyone had issues with flake8 pre-commit hooks getting confused by mypy "ignore" comments? curious if there is a recommended workaround for this aside from using --no-verify:
flake8...................................................................Failed
- hook id: flake8
- exit code: 1

pandas/core/generic.py:701:36: F821 undefined name 'ignore'
pandas/core/generic.py:701:36: F821 undefined name 'arg'
pandas/core/generic.py:4220:39: F723 syntax error in type comment 'ignore[return-value, arg-type]'
pandas/core/generic.py:4283:39: F723 syntax error in type comment 'ignore[return-value, arg-type]'
Joris Van den Bossche
@jorisvandenbossche
@dsaxton yes, I constantly have the same issue lately
quite annoying
Do you know if this is specific to pre-commit, or do you also get it when directly running flake8 ?
Joris Van den Bossche
@jorisvandenbossche
So if I just run flake8 on the terminal, I get exactly the same issue. But so not sure why we don't have it in our code_check script on CI
Joris Van den Bossche
@jorisvandenbossche
Ah, I see your PR now, thanks! Was just going to do the same ;) (still don't understand why it doesn't fail on CI though, as flake8 is pinned there at < 3.8)
Erfan Nariman
@erfannariman
@jreback looks like Series/DataFrame.slice_shift() and Series/DataFrame.tshift() are deprecated since 1.1.0, although in #18262 it's still unchecked, is there a reason for that?
Joris Van den Bossche
@jorisvandenbossche
@erfannariman slice_shift doesn't seem to be deprecated?
but tshift indeed is, updated the issue (thanks for noting!)
Erfan Nariman
@erfannariman
Yes you're right, my bad. Will see if I have time to pick up depr of slice_shift.
Daniel Saxton
@dsaxton
@jorisvandenbossche thanks for approving! that is strange about the CI itself, although i'm not sure which version specifically was when it got fixed in flake8
Joris Van den Bossche
@jorisvandenbossche
CI is using 3.7.9 (based on the output), not 3.7.7 as in the old black config, but I tried that version locally and still got the errors ...
anyway, problem solved ;)
Erfan Nariman
@erfannariman
How are we doing type annotation for sqlalchemy objects right now? I have no experience with that.
For example for a engine connection or cursor.
Daniel Saxton
@dsaxton

lots of failed CI jobs coming from Travis if anyone knows what this is about:

    @pytest.fixture()
    def gbq_dataset(self):
        # Setup Dataset
        _skip_if_no_project_id()
        _skip_if_no_private_key_path()

        dataset_id = "pydata_pandas_bq_testing_" + generate_rand_str()

>       self.client = _get_client()
...
>       if credentials and client_options.credentials_file:
E       AttributeError: 'ClientOptions' object has no attribute 'credentials_file'

https://github.com/pandas-dev/pandas/blob/c8b44dda08ab3a2feb2c4572be960469c3f065f7/pandas/tests/io/test_gbq.py
https://api.travis-ci.org/v3/job/728073325/log.txt

2 replies
Irv Lustig
@Dr-Irv
I know that as a maintainer, I can type "/azp run" to rerun the Azure tests. I'm reviewing a PR where only the Travis test failed due to a timeout. Is there an equivalent way to rerun just that test? If not, what do I tell the PR owner to do?
Joris Van den Bossche
@jorisvandenbossche
I didn't know about azp run ;)
I can restart a specific build on the travis interface
but not sure how the "rights" to do that work (not sure if that is linked to commits rights)
(now, travis is failing almost anywhere the last few days, so not necessarily useful to rerun at the moment)
Irv Lustig
@Dr-Irv
The PR is pandas-dev/pandas#36437 so if you can try on that one, it might help. Thanks @jorisvandenbossche
Joris Van den Bossche
@jorisvandenbossche
so you can't do it on travis?
image.png
for me there is both a "restart build" to restart all jobs, as a button on the left next to each job
(but restarted)
Irv Lustig
@Dr-Irv
No, I don't have that button. So I guess I don't have the permissions
Joris Van den Bossche
@jorisvandenbossche
Hmm, no idea how those permissions are managed ..
Irv Lustig
@Dr-Irv
Is it worth creating an issue?
Joris Van den Bossche
@jorisvandenbossche
maybe first check if @jreback knows about this?
Jeff Reback
@jreback
assume you logged into these via github? it should just work
Irv Lustig
@Dr-Irv
@jreback Yes, I'm logged in via GitHub, click on the Travis run in a PR, and can't restart the build
Jeff Reback
@jreback
click on your picture & settings, there is a sync account button. i think it should show the pandas in the organization
Irv Lustig
@Dr-Irv
@jreback It shows "pandas-dev" in my organization list. Maybe something has to happen with the Travis integration to sync up that I'm now on the pandas-core team??
Jeff Reback
@jreback
right did you do a 'sync now' button? then maybe log out of travis and log back in
this is all controlled via github permissions, which you have
Irv Lustig
@Dr-Irv
@jreback Thanks. Turned out that I needed to log back out/back in to travis. Now it works
Jeff Reback
@jreback
:->
Leon
@LamedB_twitter
Hey, I couldn't find a way to read parquet statistics ( Min / Max ) when I read the file from S3. I tried to covert DataSet to Table, but all the metadata is empty. How can I get it?
Irv Lustig
@Dr-Irv
@LamedB_twitter this channel is for pandas development issues. Please post your question to stackoverflow, which is where it will hopefully be answered
Leon
@LamedB_twitter
ok, sorry, thanks