On 9/15/10 1:37 AM, Julian Reschke wrote:
What browsers do today is a bug per spec.
Well, yes, but they implemented it that way as a specific workaround for broken servers, no?
Use cases for XHR are not identical to surfing web pages. There are protocols/implementations that expect method names to be preserved for 301/302 (*), and an XHR2 with sane redirect handling should at least make it possible to talk to these servers.
OK, I almost buy this argument. I feel that we should require opt-in to this behavior difference, though, not do it by default.
-Boris
