[patch] unit tests showing humanize intcomma handling floats

[Chris Beaven]

The intcomma filter should handle floats, too.

Of course, the name doesn't make much sense in that case, so I have renamed to commaseparate (while keeping backwards compatibility).

[Chris Beaven]

On a related side note, there's one spelling mistake in a docstring somewhere which misspells separate as "seperate".

I can't be bothered opening a ticket for something that negligible, but feel free to fix it as part of this one ;)

[Malcolm Tredinnick]

I'm going to leave this one until after 0.96 (which will be "real soon now"), just to avoid any confusion with the renaming change (I realise this is backwards compatible; good thinking there). Between 0.96 and 1.0, the list of things people will have to look out for will grow and we'll include this in that.

[Russell Keith-Magee]

unit test for humanize.intcomma

[Russell Keith-Magee]

Are you sure this change is required? I added a unit test for float separation (diff attached), and it worked fine as-is.

Can you provide a specific test case where the existing intcomma filter fails for floats?

[Chris Beaven]

I'm quite confused also - I can see the current method works perfectly with floats so I'm not sure why I even wrote this patch.

[Chris Beaven]

Actually, the unit test may as well go in there. I opened a new ticket regarding renaming the filter to something more sane: #3884

[Malcolm Tredinnick]

(In [4880]) Fixed #3017 -- Added tests for humanize filter.

[Malcolm Tredinnick]

Russell: I changed one of the tests (the one with a long float) before committing, because it was vulnerable to internal floating-point representation problems. I only caught it because it failed for me.