User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:2.0b13pre) Gecko/20110318 Firefox/4.0b13pre
One line fix.
Created attachment 534364 [details] [diff] [review]
Comment on attachment 534364 [details] [diff] [review]
seems fine. child listeners won't be able to mess with the cookie header and have it take effect but that seems fine.
cc-ing cookies "usual suspects" in case they have opinions:
> child listeners won't be able to mess with the cookie header and have
> it take effect but that seems fine
Some more details. In the parent, on-examine-response observers are called before cookies get parsed, so they can remove/tweak cookies. This patch calls on-examine-response observers in the child after that, so they wouldn't have the same opportunity.
This could in theory be fixed, by sending IPC traffic and having the child notify the observers (and ship back any channel header changes made back to the parent) before parsing the cookies. The cost would be an IPDL round trip, which sucks a bit given that this is the uncommon case, but we'd incur it for all traffic.