Hi Tom.

Attached is a change set with actually lets you use the debugger :-). 
Remember to use this on a disposable image; the image won't be usable 
after running this.

Run "Future doing: [ByteString]" to crash the image.

Tom Phoenix wrote:
> On 6/21/06, Michael van der Gulik <squeakml at gulik.co.nz> wrote:
> 
>> I've got this really difficult bug.
> 
> I'll say; when you subclass ProtoObject, it's easy to forget about
> some vital message that's usually handled in Object.

Oh, I'm quite used to working with ProtoObject now. I've been capturing 
messages for a couple of years now and doing crazy things with them.

>> When I execute the following, Squeak crashes:
> 
>> Future doing: [ByteStream].
> 
> What's a ByteStream? Do you mean ByteString?

Ahh... yea. One of those :o).

> You can debug the debugger to find where it crashes. Inspect the
> debugger's model, then debug what happens when you send it #send.
> That's how I found that the debugger chokes for the first time when a
> MessageCapture is sent #class. If I fix that, it chokes for the second
> time when it gets #instVarAt:. I'm sure there are more after that, but
> I'll let you find them.

Yea, I forgot to include all this stuff in the previous change set.

I have a forbidding feeling that the answer will lie in the virtual 
machine implementation. Current I'm just trying to get 
InterpreterSimulator working (from VMMaker-tpr-58), but that doesn't 
seem to run. I've fixed the first few bugs, but then it wants to coerce 
a Symbol (acting as a function pointer) to a long for inclusion in a 
method cache...!?

Mikevdg.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: DPON-bug.2.cs
Type: text/x-csharp
Size: 16748 bytes
Desc: not available
Url : http://lists.squeakfoundation.org/pipermail/squeak-dev/attachments/20060622/69ccdae0/DPON-bug.2.bin