In the link checking code (see bug 91388 for first implementation), we should build a JS array of objects containing the element whose URL was checked (usually an <img> or <a>), and the result of the URL checking (whether the failure was "timeout" or the file wasn't found.) This will be passed to the Link Check dialog that will show the user which links were bad and allow them to edit the properties on the object owning the URL.

Looking at nsIURIChecker, I realized it didn't make sense to add it to that class. But in the JS, we have an nsIURIRefObject. Here's a patch to save those in a parallel array. Does this (patch coming) seem like a reasonable solution?

Good start. I actually only need the object when the link fails to load, and I need to know if the cause is a 'timeout' or failure to find the file at all.

We have to save all the ref objects, because we don't know at the time we're saving them whether their links will turn out to be live or not. The link checker object tells you whether it was a timeout or a bad link.

I believe this got checked in along with the new dialog.

yes, it was checked in, so I'll verify this