scc and I found this while fixing the memory leaks in the webShell. the doc loaders all own their parents, not the other way around as it should be. Filing a bug so that we don't forget to fix this. someone on the gecko team should own this, but in the meantime...

This is really a design inconsistency rather than a bug. I would like to discuss it with Kipp and Troy before arbitrarily re-arranging it to fit my model of the universe. This code is currently working, it's just that the design is upside- down. Perhaps they have a really good reason for making it be that way and the only fix required is a comment. I'm moving this bug out for their input.

Actually, the allocation/de-allocation bug (bug #9341) is significantly more important than this bug, so I'm moving this bug to M9, to focus on memory problems

Well ... it's got its own ownership model, and it's working. I think this is best left alone

Based on the comments below, marking as verifed fixed.