HandleBeforeWindowClose and HandleLoadEnd

[emerick]

I've run into situations (typically when loading long web pages) where HandleLoadEnd is being called after HandleBeforeWindowClose. The situation usually looks like this:

1. Load a long-ish web page (maybe NYT front page or Google News or something like that).
2. Close the browser before it finishes loading.
3. HandleBeforeWindowClose triggers.
4. HandleLoadEnd triggers.

In my particular application, I end up crashing in HandleLoadEnd (step 4), because I'm trying to dereference a cached browser pointer that was set to NULL by HandleBeforeWindowClose. Obviously, I can guard against dereferencing the NULL pointer, but I'm wondering if this sequence of events is intended. It seems to me that when HandleBeforeWindowClose triggers, I shouldn't receive any additional events from CEF after that point, since that's occuring post-WM_DESTROY.

Please correct me if I'm mistaken in any of this! If this is indeed a bug, I'd be happy to open a new case; I just wanted to run it by everyone here first.

Thanks,

Emerick

[magreenblatt]

HandleBeforeWindowClose is called before the WebKit window is destroyed. If a page is currently loading then that load is not canceled and HandleLoadEnd is not called until the window is actually in the process of being destroyed. This is likely the situation that you're seeing.

It would probably be better to cancel the load and make sure HandleLoadEnd is called before HandleBeforeWindowClose.

[emerick]

Right, I think that makes sense. Would you like me to enter a bug for this?

Emerick

[magreenblatt]

Would you like me to enter a bug for this?

Sure. A patch would be even better .

[emerick]

Haha, fair enough. I'll start by entering a bug.