Why not inherit these from axes.adjustable or axes.anchor? A lot of other 3D properties fall back on this without setting an explicit 3D param.

#25641 :-)

One can for sure discuss which can be the same and which are beneficial to have separately.

#25641 is definitely interesting.

For this one, as defaults, maybe defaulting to what is done is 2d is useful?

Sorry, I do not get it. Are you saying that if the 3D-version of the RCparam is not changed, but the 2D-version is, it should use the 2D-version? How do we know if the style-sheet has set the 3D-version to the default value or not? I may be missing something, but I didn't think it was possible to query if the param comes from the style sheet or if the default is used?

Also, wouldn't that be a bit confusing? Now we add double parameters that are also physically located quite nearby, so I guess if someone want to change the 2D-case, chances are that they will notice the 3D-case as well?

Then it may very well be that it is not worthwhile to have separate 2D and 3D versions for all parameters introduced here. aspect and ´adjustableseems to make sense to have separately, which leavesanchor`. That was basically just added because it ended up nearby in the code and I realized that it could be done.

So, I found a few places where we do this

wherein a 3d parameter is set on based on the value of a "2d rcparam", which is why I thought we can use mpl.rcParams['axes.anchor'] directly.

Now, I personally would prefer more rcParams instead of less with inheritance as you mentioned, but I brought it up because there were some such examples. Am not opposed to adding the new rcParams though.

OK, then I understand. Yes, it is not a problem as such, more a question of which parameters we would like to separate.

Yes. But I guess that's a part of a bigger conversation about the role of rcparams itself. For now, I don't mind having all 3 separate parameters either.