[webkit-reviews] review granted: [Bug 83918] REGRESSION (r100311): YummySoup app crashes when trying to print : [Attachment 137431] patch

[bugzilla-daemon at webkit.org]

Alexey Proskuryakov <ap at webkit.org> has granted Nate Chapin
<japhet at chromium.org>'s request for review:
Bug 83918: REGRESSION (r100311): YummySoup app crashes when trying to print
https://bugs.webkit.org/show_bug.cgi?id=83918

Attachment 137431: patch
https://bugs.webkit.org/attachment.cgi?id=137431&action=review

------- Additional Comments from Alexey Proskuryakov <ap at webkit.org>
View in context: https://bugs.webkit.org/attachment.cgi?id=137431&action=review


> Source/WebCore/loader/SubresourceLoader.cpp:318
>	   m_resource->stopLoading();
>	   m_documentLoader->removeSubresourceLoader(this);

It would be nice to one day get rid of these non-"release" parts too.

> LayoutTests/http/tests/xmlhttprequest/cancel-during-failure-crash.html:17
> +img.src = "force://didFail/fromnetworkstack";

Hmm, I didn't know that we were getting a failure from network stack in this
case. Logically, WebKit may be doing sanity checks on protocols.

Approaches I've used in the past were accessing an unused port on localhost
(port 7 is one unlikely one to have listeners on), and doing an infinite
redirect (which is useful when you need to do it conditionally, but that caused
some problems with test performance on chromium IIRC). There may have been
other tricks I've forgotten about - in particular, the infinite redirect trick
in app cache directory was probably updated to something different.