[Webkit-unassigned] [Bug 38742] [chromium] Implement canEstablishDatabase call for workers.

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=38742





--- Comment #24 from Michael Nordman <michaeln at google.com>  2010-05-11 12:24:29 PST ---
> > e.g. createApplicationCacheHost should not be called either, I'd rather add ASSERT_NOT_REACHED to all methods that should not be invoked (but I cannot judge which these are), or none.
> > 
> I understand your desire for consistency. If you want to change those other
> lines to be ASSERT_NOT_REACHED() also (and run the appropriate tests to make
> sure it doesn't break anything) I'm fine with that. The worker interface is 
> confusing enough with various interfaces that are implemented but never called 
> on certain sides of the renderer/worker divide, that I feel pretty strongly
> that we should be using ASSERT_NOT_REACHED() wherever possible, even if past
> authors haven't been rigorous in this regard.

The traversing into and out of and back into (etc) in various processes really makes it challenging to grok the worker interfaces, especially when the same interfaces are used for very different purposes in different places. Interfaces like WebSharedWorker and WebWorker are difficult to get a handle on because they are both used by chrome to call into webkit and used by webkit to call out to chrome, and some of the methods are only applicable on one side or the other.

What may be nice is if the worker interfaces where more segregated than they are. One programming interface for use in the process that creates a javascript worker object. Another for use in the process that realizes the worker thread. Could possibly help to avoid NOT_REACHED methods altogether.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.