Better reverse() handling for the (.*) pattern

[Malcolm Tredinnick]

Right now, if you have (.*) in a URL pattern, calling reverse() or {% url... %} requires you to pass in an empty string argument. But this is really an optional argument, since it successfully matches nothing. It would be nice if reverse URL resolving understood that case and allowed for no parameter to be passed in.

[Malcolm Tredinnick]

Now that I look at the code, this would be unnecessarily fiddly to handle. We currently can completely ignore the contents of capturing groups and having to process them just for a few special cases would be extra work for no huge benefit.

Turns out that (.*)? would work as a pattern if somebody didn't want to have to pass in empty arguments. So wontfix-ing my own ticket, upon reflection.

[Malcolm Tredinnick]

(In [9217]) Fixed #9239 -- Somebody read my form validation docs from r9177 and found a
typo in them. I guess that was predictable. Thanks, Berry Groenendijk

[Malcolm Tredinnick]

(In [9218]) [1.0.X] Fixed #9239 -- Somebody read my form validation docs from r9180 and
found a typo in them. I guess that was predictable. Thanks, Berry Groenendijk

Backport of r9217 from trunk.

[Malcolm Tredinnick]

Above commit messages were typos. They should have referred to bug #9329.

[rdakin]

Does that mean this should still be resolved wontfix? The issue is still apparent in my testing.

[Malcolm Tredinnick]

Yes, it's still a wontfix.