[Webkit-unassigned] [Bug 137297] Allowing eval'ed code to be named in stack traces

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=137297

--- Comment #10 from Joel Einbinder <einbinder at chromium.org> ---
(In reply to Devin Rousso from comment #9)
> Comment on attachment 411597 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=411597&action=review
> 
> > Source/JavaScriptCore/runtime/StackFrame.cpp:70
> > +    String sourceURLDirective = m_codeBlock->ownerExecutable()->source().provider()->sourceURLDirective();
> 
> Are we sure that `provider()` exists?

A few lines later, ownerExecutable()->sourceURL(), calls provider()->sourceURL() internally. I can add sourceURLDirective to ScriptExecutable to make it look less sketchy. I guess I was cutting corners in pursuit of a smaller patch.

> > LayoutTests/inspector/model/remote-object/error-expected.txt:34
> > +        "_name": "sourceURL",
> > +        "_type": "string",
> > +        "_value": "__WebInspectorInternal__"
> 
> This slightly concerns me, as it exposes the fact that the evaluation came
> from Web Inspector.

I totally agree that it's concerning. However I think it just makes it more clear what's going on. The stack trace from the console is already highly detectable. It looks like

global code
evaluateWithScopeExtension@[native code]

_wrapCall

Fixing this requires stripping the stack from the console evaluations. And even then, the lack of a stack could be detected. For comparison Firefox's stack is

@debugger eval code:1:1

Whereas Chromium's is

Error
    at <anonymous>:1:1

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20201016/0e0b8abd/attachment.htm>