95534 - JSTRAP_CONTINUE doesn't

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Robert Ginda]

Build the jsdb shell, load a .js file containing the following script:
    function toCall() {
        var a = 0;
        debugger;
        var b = 20;
    }
    toCall();

When the debugger breaks, type ``show("c+d")''

You'll crash in jsinterp.c, trying to add two bogus slots.

Brendan discovered the problem to be that the JSTRAP_CONTINUE case of the switch
at line 3730 of jsinterp.c blows off the exception and sets ok to true, instead
of letting the exception bubble (as I believe it should.)

Patch coming up.

Comment 1

[Robert Ginda]

Attachment: patch, round one

Comment 2

[Robert Ginda]

r=rginda
jband/shaver, care sr= this?

Comment 3

[Brendan Eich [:brendan]]

Why did that goto no_catch; line diff?  Oh, you added trailing whitespace to the
line under JSTRAP_ERROR -- please fix and r=[e-mail].  Shaver'll sr,
I'm sure.

/be 

Comment 4

[John Bandhauer]

My original intention was to do what the code looks like it is trying to 
do: allow the debugger to detect the exception and tell the engine to just 
pretend that an execption was not raised. I'll buy the idea that this is not a 
reasonable thing to expect the engine to be able to do - since the code raising 
the exception may not have left things in a state where the next instruction can 
be executed as if nothing had gone wrong. So, having JSTRAP_CONTINUE mean "do 
what you would have done if the hook had not been called" is the Right Thing.

sr=jband

Comment 5

[John Bandhauer]

Oh, and thanks for fixing my broken code!

Comment 6

[Robert Ginda]

extra whitespace removed, fix checked in.