whoaboy, this is bad.
The problem here is illustrated in bug 687710 comment 41.
Created attachment 570910 [details] [diff] [review]
I thought about writing a test for this, but I'm not sure if there are web-visible effects of this change. It depends on who reads the parent SHEntry's URI, I guess...
Comment on attachment 570910 [details] [diff] [review]
Please fix the copy constructor too. It needs to copy mParent, right?
r=me with that.
What about the copy constructor? See comment 3!
I fixed it! But then I didn't qref, and I pushed my patch to my inbound tree...ack.
Hopefully with less orange: https://hg.mozilla.org/integration/mozilla-inbound/rev/f3babd4e975f