Even though it tries using the calling context's error reporter, the DOM's NS_ScriptErrorReporter will fail and bail out early for want of an nsJSContext private data record for the cx -- see the XXX comment at the top. This should be easy to fix. Patch in a bit. /be
Created attachment 189848 [details] [diff] [review] proposed fix I fixed another bug: the gratuitous GC from JS_DestroyContext after each and every evalInSandbox, which was bloating up compile time horribly. /be
This may matter a lot to GreaseMonkey. /be
Status: NEW → ASSIGNED
Priority: -- → P1
Target Milestone: --- → mozilla1.8beta4
dbradley, feel free to r+ if you are around. /be
Comment on attachment 189848 [details] [diff] [review] proposed fix Triple-stamping=shaver.
Fixed. Could justify this on branches at some point, but not clear it would make GreaseMonkey's (or anyone else's) life simpler, given downrev 1.0.x clients in the field that would need a workaround (which we have, I worked it out with Aaron on IRC today -- don't ask!). /be
Status: ASSIGNED → RESOLVED
Last Resolved: 13 years ago
Resolution: --- → FIXED
The compiler complains about this code: jsapi.c:4482: warning: control reaches end of non-void function Can we have a return value here, please?
I just fixed that, sorry for missing it on review.
You need to log in before you can comment on or make changes to this bug.