[ENH] Enable full compatibility with Pylance

[GitHunter0]

Feature Type

• [ ] Adding new functionality to datar

• [X] Changing existing functionality in datar

• [ ] Removing existing functionality in datar

Problem Description

datar documentations are not fully discovered and displayed by Pylance as described in this issue https://github.com/microsoft/pylance-release/issues/4603

Feature Description

For example, being able to hover on glimpse in the MWE below and have the function info and docstring displayed:

import datar.all as d
import pandas as pd
df = pd.DataFrame({'x':[1,2,4]})
d.glimpse(df)

Additional Context

No response

[pwwang]

Should be enhanced by v0.13.0

However, this still has limitations for the dynamically added functions or objects, for example, additional functions that are added by a backend, and conflict names (python reserved names, i.e. filter). One should use filter_ instead. One could use filter with options(allow_conflict_names=True). However, the IDE will still only have hints for filter_.

[GitHunter0]

Hey @pwwang , this is great, thank you.

Just a couple of points: When installing datar 0.13., I'm getting:

ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
datar-pandas 0.3.0 requires datar<0.13,>=0.12, but you have datar 0.13.0 which is incompatible.
datar-numpy 0.2.0 requires datar<0.13,>=0.12, but you have datar 0.13.0 which is incompatible.
Successfully installed datar-0.13.0

Also, when I run this

import datar.all as d
d.filter

I'm getting this error:

AttributeError: module 'datar.all' has no attribute 'filter'

However, d.filter_ runs fine.

python      : 3.10.4 | packaged by conda-forge | (main, Mar 30 2022, 08:38:02) [MSC v.1916 64 bit (AMD64)]
datar       : 0.13.0
simplug     : 0.3.2
executing   : 1.2.0
pipda       : 0.12.0
datar-numpy : 0.2.0
numpy       : 1.23.1
datar-pandas: 0.3.0
pandas      : 1.4.3

[pwwang]

Need to upgrade the backends as well.

d.filter to raise an error is by design, see: https://pwwang.github.io/datar/options/#allow_conflict_names

[pwwang]

The backends are now upgraded.

[GitHunter0]

d.filter to raise an error is by design, see: https://pwwang.github.io/datar/options/#allow_conflict_names

Yes @pwwang , but if I'm importing datar as d and using filter as a function from source d like d.filter(), there should be no conflict of names at all, right? In this situation, options(allow_conflict_names=True) should not be necessary, don't you agree?

[pwwang]

Done with v0.13.1.

Note that d.filter still doesn't have IDE hints, as it's added dynamically.

[GitHunter0]

Note that d.filter still doesn't have IDE hints, as it's added dynamically.

It's a pity, I hope Pylance gets dynamic support some day, many packages suffer from that same issue...

Done with v0.13.1.

Thank you

[pwwang]

That's the best we can do. It's (almost) impossible to do it dynamically because that requires us to run the whole code, which will bring uncontrollable side effects.

[GitHunter0]

I understand and I appreciate the feedback, that's why I hope Pylance gets smarter.