url testing is moving along with good progress, but is clear that we need continually apply priorities to which sites we test to get the best bang for the buck. if we have some data lying around, or if its possible to record such data, it would be really handy to have the time of the url load initiation to be recorded and passed on with the crash report from the client. that way we could draw a stronger relationship with the url load and the crash. if there are many minutes between url load and crash we could lower the priority for testing since the site might have less association with the crash event. If there are only a few seconds between url load and crash there might be higher chance at identifying drive-by crashes.
I think a better approach is to test links from URLs in crash reports. This acknowledges that in crash-on-load scenarios, we usually have a previous URL, instead of just giving up on those cases.
Summary: add time of of url load to crash reporting. → add time of url load to crash reporting.
We might also just fix the code that sets the current URL to be more accurate (bug 411930). Currently it's hooked into the sessionstore code, which I'm fairly certain means that if you navigate to a new page, the URL isn't saved immediately. Your proposal is complicated by tabbed browsing. Do we store the time of load for every tab, and switch the annotation on every tab change?
if we can fix 411930 the need for this probably goes down. I think to really get an accurate picture of any urls associated with crashes we do need to swtich the annotation on every tab change.
You need to log in before you can comment on or make changes to this bug.