This PR seems to restrict the subplot params into the the range [0,1]. This not useful.
All you have to check for is that left is smaller than right and bottom is smaller than top.
In addition you may check for left==right, but that may need to be allowed too, and it may be a fall-back in tight_layout.

I don't like figure.subplot, because it sound so general... like "Hell yes, I want a subplot, so let's set it to True". But I suppose it's the natural name, given the current params.

I just copied the name from figure.subplot.foo, don't really like it either, happy to change it for something better.
The 0 < ... < 1 restriction comes from the rcParams validator, will drop it, but note (as a side point) the benefit of having a single validation point for the rcParams...

Uh, so maybe that validator should change. I agree that it doesn't make sense to be able to set values larger 1 only via the api but not via rc.

Resons to keep it as it is:

• You may simply want to have >100% space between subplots (e.g. for small plots with large captions).
• When using the bbox_inches="tight" saving option, you may use some negative margin to get a desired output.
• When combining subplots with 3d axes you might want to fiddle around with the subplot_params to get a 2d and 3d plot nicely aligned.
• When creating a wide figure with an image and a colorbar you may use left < 0 to center the image. When creating a subplot grid with colorbars you may use right > 1 to reduce space to the right of the colorbar.
• When creating a wedged polar plot, you may need out of range params to get the plot tight into the figure (Wedged polar plot's position/scaling leaves too much white space. #10264)

The latter are mainly workarounds, but as long as such problems do not have a good solution (and there will always such special cases I figure) it would be a pitty to artificially limit the available options to solve them quickly.

Oh I see, the validators for hspace and wspace were just completely wrong too... killed them as well.

Since this Pull Request has not been updated in 60 days, it has been marked "inactive." This does not mean that it will be closed, though it may be moved to a "Draft" state. This helps maintainers prioritize their reviewing efforts. You can pick the PR back up anytime - please ping us if you need a review or guidance to move the PR forward! If you do not plan on continuing the work, please let us know so that we can either find someone to take the PR over, or close it.

Just for tracking purposes, the surrounding machinery has mostly all been pr'd separately; the only question remaining is whether we want to group the rcparams into a single dict or not. (I still think this would be a good idea -- probably easiest would be to just start a new pr from scratch.)

the only question remaining is whether we want to group the rcparams into a single dict or not.

I think we should not do this right now. While a single dict would make some things easier, other things get hader. You'll have to write rcParams["figure.subplot"]["wspace"] instead of rcParams["figure.subplot.wspace"], which is more cumbersome and an API change. The dict-base rcParams is not suited to make the transition easy and it also lacks features to support all use cases well.

I suggest to defer this to the redesign of the config system #24585. In particular. we can build a proper hierarchial interface there, which supports both rcParams["figure.subplot.wspace"] and config["figure.subplot"]. With a bit of careful design, the latter could be a dict or dict-like.