When the server returns the header data for the curl backend it is passed to be converted as a latin1 string. This is not always good. For example there is a testcase which fails currently: http/tests/download/literal-utf-8.html. The solution would be to process the input headers as utf8 strings. This solution in the libsoup backend.
I mean: "This solution in the libsoup backend." -> This solution is also used in the libsoup backend.
Created attachment 203694 [details] proposed patch
Comment on attachment 203694 [details] proposed patch r=me
Comment on attachment 203694 [details] proposed patch Clearing flags on attachment: 203694 Committed r151195: <http://trac.webkit.org/changeset/151195>
All reviewed patches have been landed. Closing bug.