Bug 85200 – Change JSObject::allowsAccessFrom to take a lexical global object, not a call frame

• Component: JavaScriptCore
• Status: RESOLVED INVALID

[Andy Wingo]

The patch to be attached exorcises ExecState* from one small part of JSObject.h.  Comments welcome.

[Andy Wingo]

Created attachment 139458 [details]
Patch

[Kentaro Hara]

Comment on attachment 139458 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=139458&action=review

> Source/JavaScriptCore/ChangeLog:7
> +

Would you explain in ChangeLog why you want to replace ExecState with the lexical global object?

> Source/JavaScriptCore/ChangeLog:10
> +        (JSC::JSGlobalObject::allowsAccessFrom): Exorcise EvalState* from

Nit: EvalState* => ExecState*

[Andy Wingo]

(In reply to comment #2)
> (From update of attachment 139458 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=139458&action=review
> 
> > Source/JavaScriptCore/ChangeLog:7
> > +
> 
> Would you explain in ChangeLog why you want to replace ExecState with the lexical global object?

Sure.  For some context, https://lists.webkit.org/pipermail/squirrelfish-dev/2012-March/000208.html.

> > Source/JavaScriptCore/ChangeLog:10
> > +        (JSC::JSGlobalObject::allowsAccessFrom): Exorcise EvalState* from
> 
> Nit: EvalState* => ExecState*

Good catch, thanks.

[Andy Wingo]

You know, I don't think this patch is a good one in the end, because it's too easy to confuse the arguments.  Although this function has no business being given a pointer to the call frame, the current signature does have the advantage that it is hard to misuse.  So, closing this one as INVALID.

[Kentaro Hara]

Comment on attachment 139458 [details]
Patch

Clearing r?