Lots of mistakes, mostly because of the grammar trying to enforce the ordering of namespace vs. import vs. normal rules.
Created attachment 44158 [details]
style-queue ran check-webkit-style on attachment 44158 [details] without any errors.
Comment on attachment 44158 [details]
> + , m_allowImportRules(true)
> + , m_allowVariablesRules(true)
> + , m_allowNamespaceDeclarations(true)
Normally I'd prefer not to name data members with verb phrases. That's why we end up with names like "shouldAllow". Not sure I have better suggestions here, though.
> + m_allowImportRules = m_allowNamespaceDeclarations = m_allowVariablesRules = false;
I'd prefer to see three conventional assignment statements instead of this one-liner.
I'm not sure all the new error conditions are being logged for the inspector's benefit.
The test case doesn't seem to cover everything you changed.
Despite the minor concerns above, r=me
Landed in r51608.
It seems the background color of fast/css/variables/misplaced-import-test becomes red by this patch. Was this intentional? If so, we may need to fix the test.