id	summary	reporter	owner	description	type	status	component	version	severity	resolution	keywords	cc	stage	has_patch	needs_docs	needs_tests	needs_better_patch	easy	ui_ux
19508	Implement URL decoding according to RFC 3987	Aymeric Augustin	Loic Bistuer	"Since #5738, when Django fails to decode an URL because it isn't valid UTF-8, it returns a HTTP 400 error with no content.

It's a bit sloppy to use a protocol-level message for an application-level requirement — in other words, to reply to a well-formed HTTP request with 400 Bad Request. [http://tools.ietf.org/html/rfc3987#section-3.2 Section 3.2 of RFC 3987] proposes a solution to this problem. Basically, non-ASCII bytes that do not create a valid utf-8 sequence should remain URL-encoded. This may not be trivial to implement, but it provide better error handling and it's normalized.

With this change, non-existing but well-formed URLs will return a 404 instead of a 400. That's what people expect, as shown in the comments of #5738 and #16541.

Django builds URLs according to [http://tools.ietf.org/html/rfc3987#section-3.1 section 3.1 of RFC 3987]. With this change, URLs will round trip cleanly through the reversing / resolving (that's one of the guarantees of RFC 3987) and Django will be able to deal with legacy, non-utf-8 URLs. I pursued these goals in #19468 with a more primitive technique (depending only on the encoding) and that didn't work out.
"	Cleanup/optimization	closed	HTTP handling	dev	Normal	fixed		anubhav9042@…	Ready for checkin	1	0	0	0	0	0