Re: [MiNT] Shutdown() discussion

[Odd Skancke <ozk@atari.org>]

On Wed, 10 Dec 2003, Petr Stehlik wrote:

> On Wed, 2003-12-10 at 13:15, Odd Skancke wrote:
> > > we would be able to copy even the code that invokes the
> > > NatFeats physically (which are currently illegal opcodes) into process'
> > > private memory. So basically there would be no pointers outside of the
> > > process memory.
> >
> >  I thought only the cookie jar was moved to process private memory.
>
> No. I'd copy even the code itself that is available at the place where
> the pointers point to. So the cookie would point to private memory,
> there would be the structure with two pointers pointing again to the
> private memory. Everything perfectly VM safe, hopefully.

 Okie, more NatFeat specific code? Or how does the kernel's
cookie-to-private code knows what the different cookies mean? Better find
another scheme.

>
> > > As I explained above NatFeat cookie need not to point to global memory
> > > (thanks to the private copy of a cookie).
> >
> >  Again, am I confused?
>
> Slightly, yes. But if you re-read my post (or the important bits I left
> quoted above) you'll get it.

 Yes, confused in that I didnt think the kernel should copy stuff from
pointers it knows not what is. Unless you plan to add if (cookie.tag =
__NF) copy... to the kerne code. Now we're cooking, eh?

>
> > And I have to say that adding code to our OS to make emulators happier
> > makes me angry. So angry and scared that I dont know if I'll release oVDI.
>
> This is an interesting reaction. We have started the ARAnyM project to
> support the Atari/TOS/FreeMiNT world and to allow everyone to
> participate in the development of new exciting software by giving them
> powerful (virtual) machines for basically ZERO price. This should have
> been the original Atari motto ("Power without the price") extended to
> absolute perfection.

 Fine and dandy, but change ARAnyM instead of changing the software to
make things work.

>
> After almost three years of development of ARAnyM we face your fear.
> Why? That just does not make sense.
>
> I think that everybody here agrees that without fast machines with
> plenty of RAM and disk space you cannot really develop modern
> application software. Take just the time needed for recompiling the
> whole FreeMiNT CVS tree on a Falcon or MegaSTE, for example. Since it's
> almost impossible to buy real TOS compatible hardware that would address
> todays needs (multimedia, internet) it made perfect sense to develop a
> virtual machine like we have done (and still are doing). The other
> option was to let the platform die.

 All fine and dandy, as long as you leave the software that is to run on
the emulator alone, and make changes to the emulator instead of to the
software it should run.


>
> This question (whether ARAnyM or death) we resolved three years ago and
> I thought everybody understood it - not necessarilly agreed with it, but
> I would never expect fear as a reaction in December 2003 (ten years
> after last Atari computer was developed and sold).

 ARAnyM or death. Reading between the lines here, you could not care less
about native hardware. New hardware developments are not welcome?
ShockHorror, CT60 really made it! This means that plans are to make MiNT
more and more emulation-friendly. Can you promise me that this wont
happen?



-- 
 Regards,

 Odd Skancke - ozk.atari.org - http://assemsoft.atari.org