567361 - Get rid of the remaining engine-internal uses of JSContext::globalObject

Status: RESOLVED DUPLICATE of bug 981218

(Core :: JavaScript Engine, defect)

Description

[Jason Orendorff [:jorendorff]]

Is explicit is better than implicit here? I think so. I'm not going to tackle
this, but I hope it'll inspire someone else.

The uses are:

- js_GetClassPrototype, when the scope argument is null and the JS stack is
  empty. (Note that there's a bug here where OOM from js_GetScopeChain is
  ignored.)

  This is the big one. Most of the 50 places that call js::NewObject would have
  to be changed to call something like, js::NewObjectWithScope, passing a scope
  parameter, such as JSVAL_TO_OBJECT(argv[-2]).

- js_FindClassObject - similar deal but not called as much

- JS_CloneFunctionObject, if parent is NULL and the JS stack is empty

- js_ComputeGlobalThis, but I think that's going away in bug 556277

- js_InitEthogram (a bug, I think -- oops, my review)

- There's an insidious comment in obj_create that mentions "the context's
  global object" in an insufficiently derogatory tone. FWIW that use of
  JS_GetScopeChain could probably be changed to JS_GetGlobalForScopeChain
  which stays on trace.

Comment 1

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

This got renamed to "default compartment object", and removed in bug 981218, AFAICT.