Bug 141636

Summary: Develop/Disable Caches does not properly disable disk cache
Product: WebKit Reporter: Antti Koivisto <koivisto>
Component: Page LoadingAssignee: Nobody <webkit-unassigned>
Status: RESOLVED FIXED    
Severity: Normal CC: commit-queue, mitz
Priority: P2 Keywords: InRadar
Version: 528+ (Nightly build)   
Hardware: Unspecified   
OS: Unspecified   
Attachments:
Description Flags
patch
none
with test none

Description Antti Koivisto 2015-02-16 01:36:15 PST
This uses cache model which is not passed to the network process.
Comment 1 Antti Koivisto 2015-02-16 01:37:52 PST
<rdar://problem/19786151>
Comment 2 Antti Koivisto 2015-02-16 01:58:47 PST
Created attachment 246639 [details]
patch
Comment 3 Antti Koivisto 2015-02-16 05:20:27 PST
Created attachment 246648 [details]
with test
Comment 4 WebKit Commit Bot 2015-02-16 09:17:49 PST
Comment on attachment 246648 [details]
with test

Clearing flags on attachment: 246648

Committed r180149: <http://trac.webkit.org/changeset/180149>
Comment 5 WebKit Commit Bot 2015-02-16 09:17:53 PST
All reviewed patches have been landed.  Closing bug.
Comment 6 Alexey Proskuryakov 2015-02-16 11:13:48 PST
Comment on attachment 246648 [details]
with test

View in context: https://bugs.webkit.org/attachment.cgi?id=246648&action=review

> LayoutTests/http/tests/cache/resources/cache-test.js:104
> +        }, 100);

This is unlikely to be the right thing. Nothing in the tests can ever be guaranteed to take 100ms at most. On a busy machine, and with hyperthreading, there are a lot longer delays.
Comment 7 Antti Koivisto 2015-02-16 11:27:39 PST
(In reply to comment #6)
> > LayoutTests/http/tests/cache/resources/cache-test.js:104
> > +        }, 100);
> 
> This is unlikely to be the right thing. Nothing in the tests can ever be
> guaranteed to take 100ms at most. On a busy machine, and with
> hyperthreading, there are a lot longer delays.

There shouldn't be anything timing dependent in cache operations anymore. If this makes the test flaky then that indicates a bug.
Comment 8 Antti Koivisto 2015-02-16 11:29:51 PST
I still keep some delay so that we don't end up testing the in-memory write cache only.