SR-6405: URLRequest: Implement normalization for httpMethod #1339
Conversation
ianpartridge
changed the title
|
@swift-ci please test |
|
I'm not particularly happy with this change, as it seems to be based on an implementation assumption that it outside of the specification of this method. Would like @phausler to opine in case there's some other assumption about behaviour here, such as whether there's a normalised cache of httpMethods. |
|
NB I'm not disagreeing that this is the behaviour on Darwin - and in fact, there are several places in the code where it only works if the httpMethod is e.g. GET instead of gEt - but I'd like to try and understand what the correct behaviour should be implemented as. |
|
This PR matches what I reverse engineered from Darwin. I'm sure there are other aspects of normalization that aren't covered here (e.g. HTTP header names) but those can wait for another day... |
|
OK, having looked into the way it works on Darwin, I agree that it does do a resolution based on a routine in CFNetwork, and then uses those in place of the supplied one if it finds them. So I withdraw my objections :) |
|
@swift-ci please test and merge |
Darwin performs normalization of
httpMethodfor 6 specific HTTP methods. Match this behaviour.