DOC: No links to prior versions docs

[DrGFreeman]

• I have checked that the issue still exists on the latest versions of the docs on master here

Location of the documentation

https://pandas.pydata.org/docs/index.html

Documentation problem

It is difficult to access the documentation of prior pandas versions. https://pandas.pydata.org/docs/index.html has no links to prior versions docs or they are difficult to find.

When working with legacy code using older pandas versions, access to the docs of a specific version, especially the API reference, can be useful.

While it if possible to find the URL, e.g. https://pandas.pydata.org/pandas-docs/version/0.24.0/ with a bit of trial and error, a list of links to the prior versions docs would be useful.

Suggested fix for documentation

Add a page in the docs with a list of links to prior versions docs.

[DrGFreeman]

I'm interested in contributing a PR to add the links on a new page, or elsewhere, if this idea is accepted by the maintainers. Feedback welcome. Thx.

[mroeschke]

Thanks for the suggestion. Since this suggestion overlaps with #42722, I am going to close this issue in order to consolidate discussion and effort towards that issue. Still happy to have your contribution towards this suggestion!

[DrGFreeman]

@mroeschke, I agree this overlaps with #42722, but I believe the intent and solutions to the two issues are different.

A solution to #42722 would add the current version number in the page header but would not provide access to the documentation of previous versions.

In opposition, a solution to this issue would add information on how to get documentation of previous versions (example below), without information on the current version (except on the index page).

Adding a dropdown menu to select the docs version would be a solution to both issues. However, the implementation of such a solution is significantly more complex and, as stated in #42722, probably not worth the effort:

Python also offers a drop down menu to select a specific docs version. I think this is overhead and not needed for pandas.

For the above reasons, I believe the two issues should be dealt with separately. If you agree, I could submit a PR that addresses the current issue with a change as in the above example.

[MarcoGorelli]

If you agree, I could submit a PR that addresses the current issue with a change as in the above example.

sounds good, feel free to open a PR