<link rel="canonical"> in 3.11 docs points at docs.python.org/3/ which is older

[ambv]

Example:
https://docs.python.org/3.11/whatsnew/3.11.html

In HTML source, you'll find:

<link rel="canonical" href="https://docs.python.org/3/whatsnew/3.11.html" />

This is a bad canonical link (404) since /3/ at the moment points at 3.9 (and very soon will point at 3.10) where there is no "What's New in Python 3.11" page. This is a problem because search engines and social networking link scrapers redirect to canonical links by default.

We should be smarter about this so that newer documentation editions than the default /3/ should simply skip the canonical link altogether. This is the simplest thing we can do that will ensure that:

• there is no 404 redirect
• nothing breaks when that version becomes the default and we start redirecting it to /3/

[ned-deily]

@JulienPalard, ideas?

[JulienPalard]

docsbuild scripts knows which version is pointed to by /3/ so it knows which version should not have the canonical link.

But I think we should keep canonical links from /anything_newer_than_slash_3/ to /3/ when the page exist to avoid crawlers to see duplicate content.

Maybe docsbuild script can be smart enough to remove canonicals to non-existing pages.

[JulienPalard]

Fixed in python/docsbuild-scripts@40ce6d1.

[hugovk]

Confirmed that:

• https://docs.python.org/3.11/whatsnew/3.11.html has no canonical link
• https://docs.python.org/3.11/whatsnew/3.10.html does:

<link rel="canonical" href="https://docs.python.org/3/whatsnew/3.10.html" />

Can we close this issue? (I would but triagers don't have close permission for this repo.)

[Mariatta]

Thanks everyone.