✔️ Linting Passed

All linting checks passed. Your pull request is in excellent shape! ☀️

_{Generated for commit: c61bbca. Link to the linter CI: here}

So as feared 1.5.X CI is not fully functional:

• build error for CircleCI likely CI Limit ninja number of parallel jobs in CircleCI #30333 needs to be back-ported (do it in this PR for simplicity)
• fetch_california_housing is giving HTTP 403, retriggering a build may get rid of it, if not we can probably work-around this

Do you know how to add just one more commit to this branch?

I used git cherry-pick 90a51c1 to get #30333 which seems to have worked but the commit hash changed :-/

I used git cherry-pick 90a51c1 to get #30333 which seems to have worked but the commit hash changed :-/

I would have done the same, so seems good enough 😉. I think it's expected that the hash changes (maybe because the timestamp of the commit changes or it may even depend on the previous commit not 100% sure). The CircleCI doc build manages to build so I think this should be alright.

I pinged Jeremie to see if he has some insight on the cerry-picking. I'd like to avoid making future updates to 1.5.X harder because I did something weird (cherrypick).

I pinged Jeremie to see if he has some insight on the cerry-picking. I'd like to avoid making future updates to 1.5.X harder because I did something weird (cherrypick).

I am reasonably confident this is the "right way" (or at least "one right way") but let's see what he says 😉

Also thinking about it, I am not sure we want to back-port the improvement to the conda-forge install instructions to the 1.5.X branch.

Not that many people go to the old versions doc pages (old means <= 1.5 at the time of writing since latest release is 1.6), and hopefully the rel="canonical" will work and search engines will stop pointing people to old versions doc pages.

I pinged Jeremie to see if he has some insight on the cerry-picking. I'd like to avoid making future updates to 1.5.X harder because I did something weird (cherrypick).

I am reasonably confident this is the "right way" (or at least "one right way") but let's see what he says 😉

That's how we proceed usually. It's expected that the hash changes, don't worry :)

Also thinking about it, I am not sure we want to back-port the improvement to the conda-forge install instructions to the 1.5.X branch.

Now that it's there it's less effort to keep it. I let you decide :)

OK let's merge this then! Thanks @jeremiedbb for the reassurance 😉

To make sure this gets deployed to 1.5.X branch we want to make sure this build is green 🤞

The build was green and rel="canonical" has reached the 1.5 website: https://scikit-learn.org/1.5/

Now we just have to wait and see whether that actually fixes the root issue, but I am somewhat optimistic 💪