400031 - Fix for Bug 352319 broken when context has debugger attached.

Status: RESOLVED FIXED

(Rhino Graveyard :: Core, defect)

Description

[Kevin Ruland]

User-Agent:       Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; .NET CLR 1.1.4322; .NET CLR 2.0.50727)
Build Identifier: 1.6R7

I took the example code attached to Bug 352319 and attached a trivial debugger to the context.  When I execute the example code it blows up in Interpreter.enterFrame.


Reproducible: Always

Steps to Reproduce:
1.
2.
3.

Comment 1

[Kevin Ruland]

Attachment: Source to reproduce the bug.

This is exactly the source from Bug 352319 except for the addition of a simple debugger to the context.  The debugger source is included in the tgz.

Comment 2

[Kevin Ruland]

I have played around with Rhino 1.6R7 and found the following code change to Interpreter.java fixes the problem.  Unfortunately, I don't have a deep enough understanding if this patch is correct or if it breaks anything else.

*** rhino1_6R7-rlg/src/org/mozilla/javascript/Interpreter.java	2008-02-06 16:05:49.000000000 -0600
--- rhino1_6R7/src/org/mozilla/javascript/Interpreter.java	2008-02-06 16:16:41.000000000 -0600
***************
*** 4074,4080 ****
                  // block ("catch" implicitly uses NativeWith to create a scope 
                  // to expose the exception variable).
                  for(;;) {
!                     if(scope instanceof NativeCall || scope instanceof NativeObject) {
                          break;
                      } else {
                          scope = scope.getParentScope();
--- 4074,4080 ----
                  // block ("catch" implicitly uses NativeWith to create a scope 
                  // to expose the exception variable).
                  for(;;) {
!                     if(scope instanceof NativeCall) {
                          break;
                      } else {
                          scope = scope.getParentScope();

Comment 3

[Norris Boyd]

[e-mail]

Attila, can you comment on the proposed patch?

Comment 4

[Attila Szegedi]

I'll need to spend some time on understanding it -- I myself only have incomplete understanding of how interpreter code works, so I'll want to stare a bit at it; probably tomorrow morning (europaish time)

Comment 5

[Attila Szegedi]

I'm looking into this now.

Comment 6

[Attila Szegedi]

Committed a (slightly  modified) patch -- looping just while we're encountering NativeWith. The problem seemed to be that top-level invocation of a Script object (which is also a function) won't get a NativeCall object even when it should be activatable (i.e. because a DebugFrame is created by a debugger for it, or because of a top-level try/catch or with statement.

    Checking in src/org/mozilla/javascript/Interpreter.java;
    /cvsroot/mozilla/js/rhino/src/org/mozilla/javascript/Interpreter.java,v  <--  Interpreter.java
    new revision: 1.342; previous revision: 1.341