Bug 135304

Summary: [iOS] REGRESSION(r171526): Images fail to load sometimes in Apple Store app
Product: WebKit Reporter: Pratik Solanki <psolanki>
Component: New BugsAssignee: Pratik Solanki <psolanki>
Status: RESOLVED FIXED    
Severity: Normal CC: darin, psolanki
Priority: P2    
Version: 528+ (Nightly build)   
Hardware: Unspecified   
OS: Unspecified   
Attachments:
Description Flags
Patch ap: review+

Description Pratik Solanki 2014-07-25 14:34:54 PDT
[iOS] REGRESSION(r171526): Images fail to load sometimes
Comment 1 Pratik Solanki 2014-07-25 14:37:53 PDT
Created attachment 235545 [details]
Patch
Comment 2 Alexey Proskuryakov 2014-07-25 14:46:27 PDT
Comment on attachment 235545 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=235545&action=review

> Source/WebCore/ChangeLog:15
> +        No new tests because no functional changes.

This doesn't seem like an accurate description - the fix is supposed to make it so that images always load.
Comment 3 Pratik Solanki 2014-07-25 14:53:51 PDT
Committed r171619: <http://trac.webkit.org/changeset/171619>
Comment 4 Darin Adler 2014-07-25 21:48:08 PDT
Check-in comment says this is “covered by existing tests”, but clearly that’s wrong. Was there a test failing when this was broken?
Comment 5 Pratik Solanki 2014-07-26 07:34:18 PDT
(In reply to comment #4)
> Check-in comment says this is “covered by existing tests”, but clearly that’s wrong. Was there a test failing when this was broken?

A number of layout tests in fast/images fail without this change. They only fail in WK1 since WK2 with network process does not go through this code path (if you disable network process you will likely hit this in WK2 as well). This is why I didn't see this bug during my original testing of r171526 and which is why this showed up as broken images in Apple Store app. Any image that we get in a single CFDataRef from CFNetwork will fail to load. This includes images that are disk cache file backed.
Comment 6 Darin Adler 2014-07-26 16:04:28 PDT
(In reply to comment #5)
> A number of layout tests in fast/images fail without this change.

OK. Thanks. I didn’t know that.