[Webkit-unassigned] [Bug 153864] Invoking super()/super inside of the eval should not lead to SyntaxError

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=153864

--- Comment #9 from Saam Barati <sbarati at apple.com> ---
(In reply to comment #8)
> Comment on attachment 274093 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=274093&action=review
> 
> >> Source/JavaScriptCore/ChangeLog:3
> >> +        Invoking super()/super inside of the eval should not lead to SyntaxError
> > 
> > Do we have a bug open for new.target inside eval?
> 
> Yes, I did this recently
>  https://bugs.webkit.org/show_bug.cgi?id=155545
> 
> >>> Source/JavaScriptCore/bytecompiler/BytecodeGenerator.cpp:643
> >>> +    if (needsToUpdateArrowFunctionContext() && !codeBlock->isArrowFunctionContext() && !isDerivedConstructorContext()) {
> >> 
> >> Why the "!isDerivedConstructorContext()" check?
> > 
> > Ah, I guess it's that way so we don't recreate a scope. I think I understand.
> 
> Yes, that is correct. It is for this case: 
> class C {};
> class D extends C {
>   constructor() {
>     eval("(()=>super())()");//Error
>   }
> }
> new D();
> Otherwise we will have two context scope for 'this'. I've spend last three
> evenings to find out why mention simple test case did not work
> 
> >> Source/JavaScriptCore/bytecompiler/BytecodeGenerator.cpp:4145
> >> +    return m_scopeNode->doAnyInnerArrowFunctionsUseNewTarget() || m_scopeNode->doAnyInnerArrowFunctionsUseSuperCall() || m_scopeNode->doAnyInnerArrowFunctionsUseEval() || m_codeBlock->usesEval();
> > 
> > I believe m_codeBlock->usesEval() should always be true if m_scopeNode->doAnyInnerAroowFunctionsUseEval().
> > You should double check though.
> 
> It is true, but unfortunately this patch cover cases when we do not have
> arrow function, for instance from previous comments, so I need to check
> m_codeBlock->usesEval(). Possible we need to change name of the function
> isNewTargetUsedInInnerArrowFunction ->
> isNewTargetUsedInInnerArrowFunctionOrEval

What I mean is shouldn't it be sufficient to just check m_codeBlock->usesEval()?
It should never be the case that m_scopeNode->innerArrowEval() is true but m_codeBlock->usesEval() is false.
Therefore, m_codeBlock->usesEval() should cover everything.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.webkit.org/pipermail/webkit-unassigned/attachments/20160316/e60eddbe/attachment-0001.html>