BUG: GH34677 Roll forward asv benchmark environment to py3.8

[matteosantama]

Partially closes #34677

[WillAyd]

I think we could also just remove this entry altogether

[matteosantama]

Just tested and that works for me as well. You think that's the better route to take?

[WillAyd]

Yea I think so

[TomAugspurger]

We need to verify that changing this won't cause a break in the timeseries at https://pandas.pydata.org/speed/pandas/.

[matteosantama]

How do we do that? I have another PR pending that needs benchmarking

[TomAugspurger]

I'm not sure what conditions cause asv to consider it a different environment.

I have another PR pending that needs benchmarking

Hopefully that isn't blocked by this issue. You can make the changes locally and run the ASVs.

[matteosantama]

Oh duh yeah thanks

[jorisvandenbossche]

I think this will cause a break in the timeseries (in a similar way as the cython version pin caused a new timeseries, see the two colors at eg https://pandas.pydata.org/speed/pandas/#algorithms.Hashing.time_frame), but I don't think there is a way around that? We need to do this update at some point, so the question is maybe if we are ok with doing it now.

[WillAyd]

Even if this breaks the time series I would be OK with doing it. Maybe after the next release so we get a clean delineation?

[jorisvandenbossche]

Yes, I think waiting until after the release sounds as a good idea