[Webkit-unassigned] [Bug 30322] WebKit level persistent caching

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=30322

--- Comment #36 from Gavin Barraclough <barraclough at apple.com> ---
(In reply to comment #34)
> > It will probably be random eviction at least initially. Naive LRU is generally a bad strategy for caches.
> 
> Won't this have repercussions for correctness and performance testing?

Interesting question. Obviously these problems already exist (since the networking layers beneath us currently likely have caches, and their cache eviction policies are presumably outside of our control). Currently I don't think we do anything to try to specify a particular cache eviction policy during testing - but it seems there is no reason why we could not do so. It certainly seems possible that we could add a different caching mode for testing purposes - e.g. unrestricted cache size, LRU, or pseudo-random with fixed seed.

The most important configuration to test is probably the one we plan to ship enabled by default, so initially the right thing to do here is probably just to determine the cache eviction policy we intend to actually use (likely random), and then test against that. But gaining the ability to control the eviction policy certainly seems like a nice side effect of moving persistent caching into WebKit; it definitely seems worth keeping in mind the possibility of taking advantage of this in the future.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20141110/0b6e4994/attachment-0002.html>