Based on the rsplit documentation, I'd expect

  'foo bar'.rsplit(maxsplit=1)

to work.  This is probably a much bigger problem than just rsplit, i.e. I doubt there is a policy about whether documented parameter names need to be usable as keywords, and if not, how one finds out which keywords are supported.

This is more of a documentation issue than anything else. The fact that the argument is named "maxsplit" and accepts a default value doesn't mean it's usable as a keyword argument. It would be difficult to document a parameter without giving it a name anyway...

As a rule of thumb, methods accepting 2 parameters or less often don't accept any keyword arguments, since there's no point in doing so.

I suggest closing this issue.

That's interesting: do we have a place where we explain how to read the doc? I mean, a place were we can provide example/explain how we write docs, so f.e.:

  str.rsplit([sep[, maxsplit]])  

is a description for a method that could accept 2 optional arguments, none of them keyargs or

  str.encode(encoding="utf-8", errors="strict")

is a description for a method that could take 2 optional args, and they are also keyargs (yeah, I know it's kinda bad-worded, but just to give the idea).

If you think it's overkill, let's just close this bug (after all, we have all the doc written our way ;)).

Cheers,
Sandro

The description for the function directive is at http://docs.python.org/dev/documenting/markup.html  However, I’m not sure whether the doc is accurate, since there was a switch from spam([style]) to spam(style=None) some time ago, for example in r73291.

Let’s continue the discussion on the other bug.