
Originally Posted by
wqweto
IMO the query string should not be escaped/decoded. My observations are that WinHttp converts query strings to utf-8 and encodes every byte with highest bit set using %XY in hex to be able to "smuggle" headers through 7-bit networks probably (no idea actually).
Perhaps I've misunderstood - but I think we have to be prepared to decode the query string (and the rest of the URL) server-side, because most browsers will encode the path/query portion of the URL. For example, this (insane) URL entered into the address bar (in Firefox):
Code:
http://127.0.0.1:8080/"test"?x=ABC DEF+"12345"
Gets sent to the server encoded:
Code:
GET /%22test%22?x=ABC%20DEF+%2212345%22 HTTP/1.1
Are you suggesting the the server should decode it, but the application on the server-side should if it wants to?

Originally Posted by
wqweto
I think URLs are a big mess already,
Ain't that the truth 

Originally Posted by
wqweto
probably would be best to make possible to submit to the http server *any* Unicode string in the query string with high fidelity i.e. convert to utf-8 and don't massage/process it too much.
Sounds like the new URLRaw property will do that, and allow the server-side app to deal with URLs (exactly as passed) however they see fit.