Keep in mind that responding to the user interrupt key can be timing dependent, so differences between the behavior of Cog and interpreter VMs may be nothing more than the effect of a faster versus slower VM running the same image. It does not mean that Cog is more (or less) error prone, it might be just that Cog is arriving at the error condition faster. Basically, the VM will try to figure out what process was *probably* running at the time of the user interrupt (but this is really only a guess as to which process the user intended to interrupt), and the image needs to figure out what process to interrupt in a debugger based on whatever it can learn out from the VM. This is not an entirely reliable operation ;) Dave On Wed, Jan 23, 2013 at 01:17:18AM +0100, Camillo Bruni wrote:

have you tried an older StackVM, I remember getting this behavior with more certainty under the CogVM...

On 2013-01-22, at 18:51, "Sean P. DeNigris" <sean@clipperadams.com> wrote:

...

Has anyone else noticed this? Open a fresh 2.0482 latest and hit Cmd-. a bunch of times. If the debugger opens on the first, close it and keep pressing the shortcut. Most presses seem to do nothing.

-- View this message in context: http://forum.world.st/User-Interrupt-Broken-tp4664619.html Sent from the Pharo Smalltalk mailing list archive at Nabble.com.