All users were logged out of Bugzilla on October 13th, 2018
Bug 1070992 left the ISO-8859-1 Gecko-canonical name and the encoder around. AFAICT, we aren't actually relying on the encoder behavior being different from windows-1252 anywhere in m-c. We do have plenty of references to ISO-8859-1 that would need updating if ISO-8859-1 stopped being a Gecko-canonical name. We should try to remove ISO-8859-1 as a Gecko-canonical encoding. However, not doing it just yet, because it would rot my queue of other uconv removals.
HTTP and IDL's ByteString (used by XMLHttpRequest, Fetch) rely on the behavior that iso-8859-1 has. Perhaps we implemented that as a simple conversion function though...
We don't use the intl stuff for ByteString. It wouldn't have the right behavior anyway, I expect, and would make sane error-reporting difficult even if we managed to get the right behavior out of it.
I'm pretty sure our networking code for HTTP headers ends up using the XPCOM string conversions that drop the high byte from char16_t and zero-extend char.
Fixed by bug 1261841.
Assignee: nobody → hsivonen
Status: NEW → RESOLVED
Last Resolved: a year ago
Resolution: --- → FIXED
Whiteboard: [fixed by encoding_rs]
Target Milestone: --- → mozilla56
You need to log in before you can comment on or make changes to this bug.