pydata/pandas

For questions about contributing to pandas, please ask in the pandas-dev/community channel

William Ayd

@WillAyd

@datapythonista or anyone else with knowledge, is there a particular reason why we customize isort to control the placement of pandas.core?

Is it purely a stylistic concern or was it driven out of circular imports

Just came to mind while reviewing #27552

Peter M. Landwehr

@pmlandwehr

Curious: does anyone have a good trivial example of an extension array for holding a fixed length numpy vector?

Jeff Reback

@jreback

@WillAyd import ordering is meant to be in ordered blocks@for consistency; it’s stylistic

William Ayd

@WillAyd

Yea isn’t that the standard isort behavior?

So we override that?

Jeff Reback

@jreback

no it’s not standard the way we order

Peter M. Landwehr

@pmlandwehr

(We are trying to move some data around after a conversion from raw numpy, so just want to be able to carry the data in the vector without it getting mangled.)

William Ayd

@WillAyd

Right so just asking if there’s an overarching reason for the way we order

Outside of pandas if I run isort it would do something like this

from pandas.api import DataFrame
from pandas.core.dtypes.inference import is_list_like

With our current config though we get this in our code base

from pandas.core.dtypes.inference import is_list_like

from pandas.api import DataFrame

The former behaviour seems more intuitive and just readable (though subjective)

Jeff Reback

@jreback

i think brock setup some rules

Gabriel Corona

@randomstuff

@jreback, Hi! I sent you an email few days ago about a possible security issue in pandas. What's the recommended course of action? Submit a public issue ?

Tom Augspurger

@TomAugspurger

@randomstuff https://github.com/pandas-dev/pandas/security/policy

Gabriel Corona

@randomstuff

@TomAugspurger , I did not find this :)

Tom Augspurger

@TomAugspurger

It’s somewhat new :)

William Ayd

@WillAyd

Has anyone used the Stale app on GitHub before?

https://github.com/marketplace/stale

Right now we manually go through and close stale projects. If it’s pretty simple I feel like this could be an easy thing to set up and it could help keep things clean without the “personal” aspect of someone going through and closing another persons’ PR

Tom Augspurger

@TomAugspurger

I think we’ve talked about it before. I’m -0.5 on bots closing issues, just because I’ve been on the other end of issues going stale and being closed becuase the maintainer hasn’t responded yet.

William Ayd

@WillAyd

Do you think that’s a problem with pandas?

I guess the counter to that is that I feel like more often than not in pandas a maintainer is going through and pinging for an update (exceptions certainly apply)

So automating could take save a little on the review side

Tom Augspurger

@TomAugspurger

Oh, I missed that this was for PRs. Are you aware of any bots that could check with the last comment was by a pandas maintainer / requested changes? Ideally we would only auto-close when we were the last one to comment.

Kolmar Kafran

@kafran

Guys, I have a large table in a database and a DataFrame. Is it possible to use a key in the DataFrame to filter the table in the database in the SQL query?

Like an inner join between the DataFrame and the DataBase

William Ayd

@WillAyd

@TomAugspurger I don’t think so but looks like the Stale bot has a daysUntilStale and daysUntilClose, so the former can be set to notify everyone (both users and maintainers) that something has gone stale

I think helpful for both sides

William Ayd

@WillAyd

Anyone else seeing the SSL: CERTIFICATE_VERIFY_FAILED in CI for the Docs build?

Joris Van den Bossche

@jorisvandenbossche

yes, it has been failing in some PRs I saw

Tom Augspurger

@TomAugspurger

I’m not able to reproduce locally though. I think the only URL we’re fetching in that file is the file is the tips dataset from GitHub, right?

Joris Van den Bossche

@jorisvandenbossche

I am also seeing similar failures on geopandas CI tests that fetch urls

so probably a problem at Travis CI

William Ayd

@WillAyd

Failing for us on the Azure build so I don’t think just a Travis thing

William Ayd

@WillAyd

So maybe something with the cert bundle not downloading as part of CI?

Behnaz khazai

@BK9966_twitter

hi, is there anyone familiar with PyDeface ?

I would really appreciate if you can reach out, I will be active on this twitter account

Joris Van den Bossche

@jorisvandenbossche

@WillAyd it's maybe a certifi/ca-certificates update that is needed (eg on conda-forge they did an update release last night)

Loïc Estève

@lesteve

I thought the same but ended up putting the blame on openssl=1.1.1d (not 100% sure). I opened ContinuumIO/anaconda-issues#11274.

(openssl 1.1.1d is not available on conda-forge which I think is the reason why there is no problem if you install from conda-forge)

Joris Van den Bossche

@jorisvandenbossche

thanks for opening that issue!

it's indeed only happening on envs that use defaults (not conda-forge)

but not on all envs using default

The failing build has

openssl 1.1.1d h7b6447c_0

while the non-failing build from defaults has

Where communities thrive

People

Repo info

Activity