Re: [uml-devel] Make time fly with UML

[Werner A. <wa...@al...>]

Michael Richardson wrote:
>   It would have been better if you had done this as a patch on top of UML
> rather than linking it in a different way with all the __wrap hackery. Why
> not just patch the source code since you have us do "make clean" and
> recompile...  The code would have been much easier for Jeff to accept that way.

Oh, I certainly don't want Jeff to accept that code :-) The wrap
hackery simply allows me to keep the UML and umlsim source separated,
which helps me during development. Once the umlsim "core" is
reasonably stable, I'll get rid of that wrapping stuff.

>   We need to agree on an IPC mechanism as well - multiple machines need to
> skip ahead together. I would propose that this needs to be coordinated by a
> small process that simply accepts time-to-next-event inputs from each UML,
> takes the minima and advanced each machine by that amount.

Yes, and there's even more, like breakpoints and perhaps direct
packet generation. There are plenty of interesting and useful things
you can do once you control time.

- Werner

-- 
  _________________________________________________________________________
 / Werner Almesberger, Buenos Aires, Argentina         wa...@al... /
/_http://icapeople.epfl.ch/almesber/_____________________________________/