Bug 26672 – google.com/adplanner shows blank page instead of site details in "basic research'

• Status: RESOLVED
• Resolution: FIXED
• Component: JavaScriptCore

[Oliver Hunt]

<rdar://problem/6940519>

* SUMMARY
Blank page shows up instead of expected content

* STEPS TO REPRODUCE
1. Go to https://google.com/adplanner
2. Log in with any google account
3. Accept terms and continue
4. Click "begin research"
5. Click on the "facebook.com" link in the site name column

* RESULTS
A blank page shows up, rather than a page full of charts and data.

[Oliver Hunt]

Created attachment 31775 [details]
Make codegen for comma nodes iterative

[Mark Rowe (bdash)]

Comment on attachment 31775 [details]
Make codegen for comma nodes iterative

We use the spelling "initialize" rather than "initialise", as you can see from the function you renamed.  Would be good to keep the test and ChangeLog consistent with that.

[Oliver Hunt]

Created attachment 31776 [details]
appease the colonies

[Darin Adler]

Comment on attachment 31776 [details]
appease the colonies

> +        virtual bool isCommaNode() const { return true; }

This should be private. No one should ever call this function if they already have a CommaNode* and making it private would detect that mistake.

I personally think we should fix this kind of thing by making *code generation* iterative rather than preventing the tree from getting deep. It seems we still can easily have problems in deeply nested cases of other types.

r=me

[Oliver Hunt]

Committing to http://svn.webkit.org/repository/webkit/trunk ...
	M	JavaScriptCore/ChangeLog
	M	JavaScriptCore/parser/Grammar.y
	M	JavaScriptCore/parser/NodeConstructors.h
	M	JavaScriptCore/parser/Nodes.cpp
	M	JavaScriptCore/parser/Nodes.h
	M	LayoutTests/ChangeLog
	A	LayoutTests/fast/js/excessive-comma-usage-expected.txt
	A	LayoutTests/fast/js/excessive-comma-usage.html
	A	LayoutTests/fast/js/resources/excessive-comma-usage.js
Committed r45106