785174 - Implement Debugger.Object.prototype.evalInGlobalWithBindings

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Thaddee Tyl [:espadrine]]

As seen on MDN [1], Debugger.Object instances have a method called evalInGlobalWithBindings() that would be very useful for evaluating WebConsole statements, thereby fixing bug 774753.

  [1]: https://developer.mozilla.org/en-US/docs/SpiderMonkey/JS_Debugger_API_Reference/Debugger.Object#evalInGlobalWithBindings()

Comment 1

[Jim Blandy :jimb]

Attachment: Refactor DebuggerFrameEval to be useful for Debugger.Object.prototype.evalInGlobal{,WithBindings}

Comment 2

[Jim Blandy :jimb]

Attachment: Implement Debugger.Object.prototype.evalInGlobal, .evalInGlobalWithBindings, .evalInGlobalWithBindingsAndABrandySoakedMaraschinoCherry

Work in progress. Currently trips cross-compartment assertions.

Comment 3

[Jim Blandy :jimb]

I'm not sure whether this addresses bug 529474 as well or not.

Comment 4

[Thaddee Tyl [:espadrine]]

My understanding is that it would, but Firebug isn't using jsd2 yet. They're making the switch, though, so they'll benefit from it as well!

Comment 5

[Jim Blandy :jimb]

(In reply to Thaddee Tyl [:espadrine] from comment #4)
> My understanding is that it would, but Firebug isn't using jsd2 yet. They're
> making the switch, though, so they'll benefit from it as well!

Okay, with the caveat that, where bug 529474 talks about a XPCSafeJSObjectWrapper, we're handing back a Debugger.Object. Seems legit.

Comment 6

[Jim Blandy :jimb]

Attachment: Implement Debugger.Object.prototype.evalInGlobal, .evalInGlobalWithBindings, .evalInGlobalWithBindingsAndABrandySoakedMaraschinoCherry

I'd love suggestions for more tests.

Comment 7

[Jim Blandy :jimb]

Try: https://tbpl.mozilla.org/?tree=Try&rev=3342a1cd7f22

Comment 9

[Jason Orendorff [:jorendorff]]

Comment on attachment 658547 [details] [diff] [review]
Refactor DebuggerFrameEval to be useful for Debugger.Object.prototype.evalInGlobal{,WithBindings}

Review of attachment 658547 [details] [diff] [review]:
-----------------------------------------------------------------

Very nice.

::: js/src/vm/Debugger.cpp
@@ +3427,3 @@
>  static JSBool
> +DebuggerGenericEval(JSContext *cx, const char *fullMethodName,
> +                    Value *code, Value *bindings, Value *vp,

Why does `code` have to be a pointer?  const Value & should work, or ValueHandle (but that's more bother for callers).

Comment 10

[Jim Blandy (bug-watching account --- use [e-mail] instead)]

(In reply to Jason Orendorff [:jorendorff] from comment #9)
> Why does `code` have to be a pointer?  const Value & should work, or
> ValueHandle (but that's more bother for callers).

There's no good reason. I just consistently forget that const Value & isn't just as relocating-GC-friendly as a handle.

Comment 11

[Jim Blandy (bug-watching account --- use [e-mail] instead)]

Er, "is".

Comment 12

[Jim Blandy :jimb]

Test comment.

Comment 13

[Jim Blandy :jimb]

I changed DebuggerGenericEval's 'code' argument to be a const Value &, and changed RequireGlobalObject to take a const Value &dbgobj, instead of taking args and calling args.thisv(), because, jeez.

Comment 14

[Jim Blandy :jimb]

https://hg.mozilla.org/integration/mozilla-inbound/rev/4afed1cd79a6
https://hg.mozilla.org/integration/mozilla-inbound/rev/955f71c08edb

Comment 15

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/4afed1cd79a6
https://hg.mozilla.org/mozilla-central/rev/955f71c08edb