Content team meeting tomorrow, 1pm Pacific

Boris Zbarsky bzbarsky at mit.edu
Thu, 17 Nov 2005 16:59:20 -0600


I'm sorry I couldn't make the meeting -- stuff came up.  Comments inline below.

>> bookmarklet) bugs and how to tackle them.  We want to fix all of these 
>> ASAP, but we haven't sized the problem.

> need highest priority on two bugs that will allow getting deeper into 
> the testing and nearer to the end of the bug list
> - https://bugzilla.mozilla.org/show_bug.cgi?id=308120  boris?

In the context of XBL1 this is pretty hard to fix...  In the context of XBL2 it 
might be easier, since then we could use ownerDocument, possibly.  In any case, 
I would have to have a much better understanding of our insertion table stuff to 
do anything for this in XBL1...  That's on The List, but not at the top...

> jonas to take the initial lead on analyzing the full list of bugs to see 
> if a few deeper fixes can fix general problems

This has come up before -- one thing that would fix general issues is making XUL 
frame construction:

1)  Well-defined (in terms of explicitly knowing what DOM leads to what layout
     model)
and
2)  Following said definition.

Then we can fix actual frame impls as needed to deal with the valid frame trees.

There are also known issues with fixed positioning, etc, crashing in XUL, of 
course.  And the perennial fragility of table pseudo-frames...

> -- ask a language how trusted it is -- python will only execute if local

How are we determining "local"?  I'd suggest using a UniversalXPConnect test, 
and then we reduce to worrying about how (or if!) that gets granted outside of a 
JS context (at the moment, the only way outside of JS to have that privilege is 
to have the system subject principal).

-Boris