Well, if you are willing to only run using POSE, you could definitely add
code to POSE to send each incoming event over to your debugger, and to stop
before/after calls to the Palm OS. Since it's happening in the emulator
itself, your app should be oblivious to it. It's similar in spirit to how
Gremlins works, generating events that just appear to the Palm OS.
Neil
> Not sure what to call this, but here goes.
>
> I am one of the developers of Pocket Smalltalk, a version of Smalltalk that
> runs on Palm OS (see www.pocketsmalltalk.com). I am in the middle of
> designing a debugger for this, such that I can run the app on the device and
> debug on the pc. Currently the cycle for debugging code is as follows:
> - write code on pc
> - package on pc
> - load in poser
> - stack trace on poser
> - find code on pc
> - modify and return to step 1
>
> I would like to be able to debug and modify code at the same time.
> Currently, I have an emulator for the Smalltalk VM, but not for the Palm OS
> calls. I really do not want to have to build a Palm OS emulator as well.
>
> One of the ideas I have is to do the following, have a stub in Poser or the
> physical device. This stub would simply take all events coming in and send
> them to my debugger, where the Smalltalk emulator would kick in. The stub
> would also receive incoming calls to Palm OS functions (systraps) and
> execute them and return the result. I'm not sure I am explaining this in
> enough detail. But I basically want to run my code on the pc (not in poser
> or device) that can affect the state of poser or the physical device.
>
> I was wondering if anyone can think of anything wrong with this. Other than
> why the hell bother, or what the hell are you doing. One issue I can think
> of is that under Windows programming there are timing issues with responding
> to certain types of events. Is there the same type of issue with Palm OS ?
> If so, is there something I can use in Poser that reduces this issue ? Any
> other issues anyone can think of ?
>
> My other option is to simply have a debug version of my Smalltalk VM that
> can talk to the PC and provide variable info, provide step functionality,
> and modification of code while debugging. This would be good and all, but I
> think the stub approach would be easier to do.
>
>
>
--
Neil Rhodes
Calliope Enterprises, Inc.
1328 Clock Avenue
Redlands, CA 92374
(909) 793-5995 [EMAIL PROTECTED] fax: (909) 793-2545
