> Does that seem like a valid alternative?

Okay, nevermind that. The latest patch handles both __layer and
__styleRuler leaks, even if layout runs before attachment or after
detachment.

Basically, in those cases, it briefly hooks them up, but pre-emptively
cleans them up in finalizeLayout. It can do this (AFAIK) because, being
unattached, it doesn't have to worry about any resize events happening
that would need them in place.

When/if attach actually happens, it puts the layer/style rulers in
place more permanently, since we know detach will happen now.

Still haven't thought about the resize handlers you mentioned, Thomas.
Getting to that, but so far they don't seem to be causing leaks (with
the cases Dirk provided anyway).

Dirk, I'd appreciate a sanity check; if you could try patch set 4 and
let me know if it a) leaks and b) still works as expected, that'd be
great.

- Stephen

-- 
http://groups.google.com/group/Google-Web-Toolkit-Contributors

Reply via email to