Stephen,

Interesting idea. It could be done by having a separate process in the
debugger (since you most certainly don't want to mess up the debuggers
own TLS) which runs the simulation. Might even be reasonably efficient.

Cheers,
  - Andreas

> -----Original Message-----
> From: squeak-dev-admin at lists.squeakfoundation.org 
> [mailto:squeak-dev-admin at lists.squeakfoundation.org] On 
> Behalf Of Stephen Pair
> Sent: Monday, February 11, 2002 8:38 PM
> To: squeak-dev at lists.squeakfoundation.org
> Subject: RE: Process local variable and debugging
> 
> 
> Wouldn't it be better to clone the process local variables into the
> process used for debugging rather than hack #activeProcess (which I
> think would create more problems than it solves)?
> 
> - Stephen
> 
> > -----Original Message-----
> > From: squeak-dev-admin at lists.squeakfoundation.org 
> > [mailto:squeak-dev-admin at lists.squeakfoundation.org] On 
> > Behalf Of Andreas Raab
> > Sent: Monday, February 11, 2002 2:26 PM
> > To: squeak-dev at lists.squeakfoundation.org
> > Subject: RE: Process local variable and debugging
> > 
> > 
> > Tim,
> > 
> > None of the stuff I did (which has been a while ago) worked 
> > correctly in the debugger. Notice that just intercepting 
> > #activeProcess won't help - it may be sent during a 
> > #quickStep (e.g., a non-simulated send during simulated 
> > execution) which is _not_ run in the simulated process.
> > 
> > Cheers,
> >   - Andreas
> > 
> > > -----Original Message-----
> > > From: squeak-dev-admin at lists.squeakfoundation.org
> > > [mailto:squeak-dev-admin at lists.squeakfoundation.org] On 
> > > Behalf Of Tim Rowledge
> > > Sent: Monday, February 11, 2002 8:01 PM
> > > To: squeak-dev at lists.squeakfoundation.org
> > > Subject: RE: Process local variable and debugging
> > > 
> > > 
> > > I like the general ideas here but have a couple of queries.
> > > 
> > > First, at the moment (as of update #4652) pool variables 
> > appear to be 
> > > accessed using the pushLiteralVariable bytecodes, which 
> > does not send 
> > > #value to the association. Changing that would involve some 
> > compiler 
> > > hacking that I hope somebody else has already done. I hate to 
> > > needlessly repeat redundantly unneccessary coding.
> > > 
> > > Second, since the current system puts actual associations in
> > > the method
> > > literal frame, simply swapping an array of associations 
> doesn't seem
> > > likely to be enough on its own.
> > > 
> > > Third, I'm not yet convinced that this approach helps with 
> > my original 
> > > problem in the debugger. If the process-globals are swapped when 
> > > process are swapped (which seems pretty much required) then the
> > > debugger process
> > > will usurp the expected values of the original code. If 
> the debugger
> > > _doesn't_ do the normal thing, then it might have problems 
> > if it were
> > > expecting to use its own process-globals anywhere.
> > > 
> > > Fourth, I seem to remember reading that there are some 
> > problems with 
> > > PoolDictionaries and modules. What to do?
> > > 
> > > So, Andreas your message implies you've been making use of a
> > > facility of
> > > this sort and I'd be delighted to get hold of the relevant 
> > > code asap :-)
> > > 
> > > tim
> > > --
> > > Tim Rowledge, tim at sumeru.stanford.edu, 
> > http://sumeru.stanford.edu/tim
> > > Don't be sexist; broads hate that!
> > > 
> > > 
> > 
> > 
> > 
> > 
> 
> 
>