Closed Bug 202994 Opened 17 years ago Closed 17 years ago
This is fairly serious, as it could allow tracking of browsing in some situations. Nominating for 1.4final blocker.
Severity: minor → major
jst's got a patch. /be
Attachment #122635 - Attachment is obsolete: true
Assignee: mstoltz → jst
Status: NEW → ASSIGNED
OS: Windows XP → All
Priority: -- → P1
Hardware: PC → All
Whiteboard: [HAVE FIX]
Target Milestone: --- → mozilla1.4beta
Comment on attachment 122639 [details] [diff] [review] This should do it. Looks good to me. r=mstoltz.
Attachment #122639 - Flags: review?(mstoltz) → review+
Comment on attachment 122639 [details] [diff] [review] This should do it. Null return from JS_ValueToString indeed means an error, and errors are exceptions (except for "out of memory" -- that gets reported via the context's error reporter immediately, no exception is created for it). Do you need the nccx->SetExceptionWasThrown call and surrounding code from JS_GetPendingException down? Shouldn't the new helper JSValueToString return false or failure, and that r.v. propagate? /be
This is baiscally the same as the above, with the addition of distinguishing between OOM and an exception being thrown in JS_ValueToString(). Brendan, yes, I do need the call to SetPendingException() on the current call context, w/o that, XPConnect will go on as if no error occured. But I do not want to throw an error to the XPCOM caller in this case, doing that makes a lot of code freak out. Ideally, we could clean up this mess and do that, but I'm not up for doing that at this point.
Attachment #122639 - Attachment is obsolete: true
Comment on attachment 122797 [details] [diff] [review] Same as above, with OOM error propagation carrying forward r=mstoltz, marking sr=me. Nit: "as if the result *were* undefined" -- uphold the dying subjunctive mood! /be
Comment on attachment 122797 [details] [diff] [review] Same as above, with OOM error propagation Requesting approval to check in for 1.4final.
Attachment #122797 - Flags: approval1.4?
Comment on attachment 122797 [details] [diff] [review] Same as above, with OOM error propagation a=as (on behalf of drivers) for checkin to 1.4
Attachment #122797 - Flags: approval1.4? → approval1.4+
Status: ASSIGNED → RESOLVED
Closed: 17 years ago
Resolution: --- → FIXED
So this caused regression bug 206026. The fix for that is to revert the evaluation context change. See bug 206026 for more details.
You need to log in before you can comment on or make changes to this bug.