[webkit-reviews] review denied: [Bug 186171] CompletionHandlers should be called on their creation threads : [Attachment 341719] Patch

[bugzilla-daemon at webkit.org]

Chris Dumez <cdumez at apple.com> has denied Alex Christensen
<achristensen at apple.com>'s request for review:
Bug 186171: CompletionHandlers should be called on their creation threads
https://bugs.webkit.org/show_bug.cgi?id=186171

Attachment 341719: Patch

https://bugs.webkit.org/attachment.cgi?id=341719&action=review




--- Comment #16 from Chris Dumez <cdumez at apple.com> ---
Comment on attachment 341719
  --> https://bugs.webkit.org/attachment.cgi?id=341719
Patch

Personally, I worry about this kind of change because of WorkQueue. On Cocoa,
WorkQueues are based on dispatch_queues. When the dispatch queue is serial,
tasks dispatched on the queue are guaranteed to run sequentially, one after
another, but they are not guaranteed to run on the same thread. As a matter,
they often run on different background thread, which is why we often have to
disable threading checks in SQLDabatase. Some objects like
ResourceLoadStatisticsPersistentStore / ResourceLoadStatisticsMemoryStore are
always used from a background queue and rely extensively on CompletionHandlers.
This usage is perfectly safe and yet would trip this new ASSERTION. As a
result, I do not believe that adding such assertion is OK / feasible.