function argument variables or variable?

[avehtari]

• extract_list_of_variable_arrays() has argument named variables
• set_variables() has argument named variables
• subset_draws() has argument named variable

All these accept one or many variable names. Should these be unified to all use variable?

Currently if the argument is variables, using variable works, too, due to how R works. Would be enough to change the documentation for these?

[paul-buerkner]

I agree that variable might be more consistent, although I think I remember we had thought about it and choose plural there for a reason (but I might be wrong). In any case, I would leave them as variables now for backwards compatibility and rely on partial matching of variable names as you say. How would you update the doc for this purpose?

[avehtari]

I meant to make a PR, but accidentally pushed to master. See commit ed4a0fe, and if not happy with it, I'll revert and make a PR

Maybe we should protect master from direct commits?

[paul-buerkner]

No worries. Had to revert your commit though because this will lead to a CHECK failure due to argument-documentation mismatch. I don't mind if we can push directly to master. If something goes wrong, I can quickly fix it, as I just did.

[avehtari]

If you reverted the commit, do I now make a PR?

[paul-buerkner]

Yes, I think so.

[paul-buerkner]

Not sure what a good fix would be for this (if any). Leaving it on hold for now.

[n-kall]

To add to this, there are inconsistencies between related packages too. cmdstanr $draws() uses plural, and priorsense changed from plural to singular before I submitted 1.0 to cran, but due to some confusions it now accepts either. I don't see any simple way to unify