Should we call them something likeaxes_transform instead of the other way?

Would be fine with that too.

+1 for namingaxes_transform and similar.

In the conflict between backward compatibility and "There should be one-- and preferably only one --obvious way to do it." I propose the following:

1. The new names should be the preferred choice.
2. They should be the attributes, the old names should be made available via properties for backward compatibility.
3. All internal code should use the new names.
4. All examples should use the new names.
5. The documentation of the old names should discourage their use and point at the alternatives.
6. I agree that we should not yet deprecate the old names, but they should go in the long run (maybe 4.0ish). It's probably good to mark them as pending deprecation.

Not everything (in particular not 3, 4) has to be part of this PR, but we should discuss the further strategy here before merging this PR.

Renamed accordingly.
You get tosed through the codebase for switching to new names everywhere :-)

Can do with 3-5.

Still, I would like to have at least 2. in this PR. Changing that later would basically revert this PR.

I actually think 2 & 3 should be done together (unless you can show that there's no performance issue, given how much we access these internally).
(FWIW I also think that deprecating the old names gets a bit close to "causing gratuitious code churn" (who'd have thought I'd once say this...).)

Ok, for 2-3 I'd basically have to create a competing PR. But we should discuss and agree if we want 1-6 before starting further work.

I am +1 on 1-5 (feel free to open your own PR) and +0 on 6.

I'm far from convinced that this renaming is a good idea--that its advantages (ability to avoid camelCase) outweighs its disadvantages (more code, more explanations required, two names for exactly the same thing).

Moving to 3.1 as this will need more consideration.

Did we still want to try and discuss for 3.1?

Moving further to 3.2 because this still needs some consideration and I don't see a dire need to rush this. Feel free to re-milestone.

General agreement during call was to close this, but@timhoffm gets to reopen this if he wants :)

Let's keep this closed for now. There are enough other battles to fight.

However, I reserve the right to revisit. 😄