ENH: Distinguish widths argument in read_fwf

[erfannariman]

• closes ENH: more distinguish name for widths arugment in read_fwf #34953
• tests added / passed
• passes black pandas
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• whatsnew entry

[erfannariman]

To start the discussion about the argument names, thought I make the changes and create a PR. To make it consistent, I changed colspecs to col_specs as well.

[WillAyd]

This breaks API pretty heavily so can't just go with it in current form; would have to go through a depreciation cycle first

This is somewhat related to #22639 though I don't think col_specs is much better than colspecs so I'd be -1 on doing this given churn

[erfannariman]

This breaks API pretty heavily so can't just go with it in current form; would have to go through a depreciation cycle first

Sure, but I'm not familiar with it, is there any documentation on that which I can go through?

This is somewhat related to #22639 though I don't think col_specs is much better than colspecs so I'd be -1 on doing this given churn

In my opinion I think going with col_specs and col_widths would make more sense, purely for the consistency and the problem I mentioned in the issue, where using width instead of widths will not throw an error and the argument names are too similar. Plus using underscore for multi word variables is agreed upon in the issue you are referencing to right?

[WillAyd]

where using width instead of widths will not throw an error and the argument names are too similar.

That is unrelated to the naming convention. I think if you wanted to remove kwargs from this and replace with the appropriate name arguments that would make more sense, as there are probably only a few that should be passed from this to the TextFileReader

[erfannariman]

where using width instead of widths will not throw an error and the argument names are too similar.

That is unrelated to the naming convention. I think if you wanted to remove kwargs from this and replace with the appropriate name arguments that would make more sense, as there are probably only a few that should be passed from this to the TextFileReader

That makes sense as well. Although I would not be sure how to decide which name arguments to include / exclude, would that be all the arguments which can be found in the tests?

[WillAyd]

The best way I think is to look at the TextFileReader implementation and see what kwargs it reads that make sense for a fixed width file

…

Sent from my iPhone

On Jun 30, 2020, at 12:26 PM, Erfan Nariman ***@***.***> wrote:  where using width instead of widths will not throw an error and the argument names are too similar. That is unrelated to the naming convention. I think if you wanted to remove kwargs from this and replace with the appropriate name arguments that would make more sense, as there are probably only a few that should be passed from this to the TextFileReader That makes sense as well. Although I would not be sure how to decide which name arguments to include / exclude, would that be all the arguments which can be found in the tests? — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

[jreback]

yeah we cannot simply change these w/o a deprecation cycle, nor do i think the churn is worthile.

However as @WillAyd indicated, remove the ability to accept **kwargs would be ok

[MarcoGorelli]

Hi @erfannariman - is this still active?

[erfannariman]

Yes, will pick this up after my vacation.

[dsaxton]

@erfannariman Can you merge master to fix conflicts?