On 2012-01-14 06:29, Jason Duell wrote:
>> Right. I am saying 307 is good enough.
> I'm in agreement with Brian here until/unless I see a use case that
> seems to merit putting in the effort here. Filling in the fourth
> quadrant of a logical grid (and making redirects from foo.com to
> foo.com/) doesn't seem like enough to me personally.
Well, the effort is done; the code is there, including test cases.
There's also a cost in not doing it (for instance, time spent explaining
to people why this status code is missing).
> I'm also with Brian in not being vehemently opposed. But let's remember
> how often little "simple" changes (especially somehow ones involving the
> cache) wind up involving a lot more followup work than we initially
> expect. So many other fish to fry...
That is true. We definitively need more test cases for the network stack.
For instance, I recently noticed that the Cache-Control header field
parser is, let's phrase it politely, very naive; so is WWW-Authenticate.
These problems may affect extensibility (introduction of new directives
or auth schemes), and might affect security (for instance, Firefox will
happily extract parameters from a quoted-string value in a new extension
parameter). See <https://bugzilla.mozilla.org/show_bug.cgi?id=716346>.