see bug 82479. We made sure that we would not send the referrer from https to http but the implementation also removed the referrer in the case when the request is to the same encrypted server. This is unnecessary broad.
*** Bug 89243 has been marked as a duplicate of this bug. ***
ddrinan: you should really use SchemeIs in place of GetScheme/strcmp.
You should be using strcasecmp to compare the schemes and the hosts, since both of those are case insensitive, according to the appropriate RFCs.
adding patch keyword.
r=bbaetz. The spec says (RFC2616, 15.1.3): " Clients SHOULD NOT include a Referer header field in a (non-secure) HTTP request if the referring page was transferred with a secure protocol." Should we check ports as well, or let it through anyway?
ddrinan, sr=darin provided you fix the indentation to make it consistent with the rest of nsHttpChannel.cpp (4 spaces of indentation).
Mass assigning QA to ckritzer.
Fix checked in.
bbaetz has informed me that this bug is needed for the branch. Re-opening and adding keyword topembed.
this is a war room bug that we'ed like to get on the 0.9.2 branch
Approved for check in to the branch by verbal comment from chofmann.
Checked into the 0.9.2 branch. Marking fixed.
Did this re-break in 0.9.3? 0.9.3 on Linux (RH7.1), I'm very clearly not getting the referer (sic referrer) header when going from one https document to a linked https document on the same server.
Just to add a clarification... the problem I'm seeing is https->https, which is technically different than this bug. BUT... this worked properly in 0.9.1, so the patch for this bug may have had the unintended side effect of messing up https->https.
This fix did not make it in to 0.9.3. It's checked into the 0.9.2 branch and the trunk.
*** Bug 93310 has been marked as a duplicate of this bug. ***
*** Bug 97303 has been marked as a duplicate of this bug. ***
*** Bug 100289 has been marked as a duplicate of this bug. ***
*** Bug 103838 has been marked as a duplicate of this bug. ***
*** Bug 96912 has been marked as a duplicate of this bug. ***