Andres,
What I still do not understand is why it is necessary to change the primitive numbers, or why it is necessary to check from Smalltalk what kind of exception support is in the VM. We're just making images and VMs incompatible with each other. I think it would be much easier to make VMs polymorphic from a primitive behavior point of view, but I'll welcome any illuminating comments any of you may have.
The reason for this particular incompatibility is unawareness about the fact that Tims original EH fixes went into the SWT image in that particular way. In other words: Tim allotted a few "marker" primitives which were spread out over a number of prim numbers and (IIRC) also conflicted with some prior uses of these prim numbers. When we adopted these changes I gave those marker primitives different numbers (because I was not aware of the use in SWT which had not seen the day of the light at that time) and therefore we got into the current mess. My suggestion for the SWT image is to just change those primitive numbers (e.g., just apply the changes I sent out a while ago) since most people will have "stock" VMs.
Cheers, - Andreas