enable proper `in` checking

[Ben-Epstein]

current behavior

import vaex
df = vaex.from_arrays(
    id=vaex.vrange(0, 200_000)
)

299_999 in df.id  # True but wrong

proposed

299_999 in df.id # False

[JovanVeljanoski]

This materializes the column right? That's.. not ideal...

[Ben-Epstein]

@JovanVeljanoski after speaking with @maartenbreddels it was my understanding is that .values is 0 mem copy. https://vaexio.slack.com/archives/C017EEHSQ84/p1646850161602389

[JovanVeljanoski]

Yes because in that example, you created an in memory dataset. So your data (and that column) is already in memory. But if you read an hdf5/arrow/parquet file, you first will put it to memory (since otherwise you are reading from disk) then you gonna do the stuff that you like.

[Ben-Epstein]

@JovanVeljanoski thanks for the explanation, I didn't understand that. So is there a better approach here? Currently returning True always could lead to many side-effects for end users

[Ben-Epstein]

@JovanVeljanoski how about now?

[maartenbreddels]

I like it, I just don't know why 7 in df["id"] evaluates to True 🤷

[Ben-Epstein]

Hmm @maartenbreddels when i tested locally i did not get that behavior. Are there docs to build vaex from scratch locally?

[maartenbreddels]

https://vaex.io/docs/installing.html#for-developers should help!