[Freemint-list] Interesting github project / ressurection

[Ryan Daum]

I agree it's a very good idea. I also think there's also some small merit
to two related ideas I've explored.

  * A native, modern implementation of at least the AES part of GEM,
running natively on Posix systems rendering to framebuffer/OpenGL or to X
or Wayland. Then a translation layer to trap 'classic' 68k Atari AES calls
and turn them into 'modern' forms. It could actually fill a niche for small
embedded-friendly lightweight GUI toolkits for games, etc.
  * Running something like TOSEMU or ParaTOS overtop of a posix layer over
a realtime microkernel rather than Linux, and port it to small hardware, to
have a "TOS-like" operating system. And port it to ColdFire.

I too attempted to revive the oaesis/ovdisis sources, and attempted idea #1
based on that. I have a couple months of heavy refactoring/rewriting (in
c++) derived from that work stashed away in a private git repo. But in the
end I found the starting point in oaesis/ovdisis wasn't complete enough and
I would have been better off starting from scratch. That code never had a
complete working GEM implementation.

Ryan



On 24 January 2017 at 07:53, Vincent Rivière <vincent.riviere at freesbee.fr>
wrote:

> On 24/01/2017 09:27, Miro Kropáček a écrit :
> > as a pure coincidence I've found this on github:
> > https://github.com/e8johan/tosemu -- it's something like wine combined
> > with aranym, seems quite active.
>
> IIRC, I already spotted Johan's TOSEMU some time ago. Or maybe I confuse
> with Keli's ParaTos, I don't remember well.
>
> I see on the readme that TOSEMU was inspired by my 68Kemu. It is also
> based on the Musashi 68000 emulator (the one used by MAME).
>
> The funny fact is that my 68Kemu was actually a fork of an older project
> of mine called RunTOS. And the purpose of RunTOS was... exactly the same
> thing as TOSEMU. So finally we have 3 projects with the same purpose:
> RunTOS, TOSEMU and ParaTos. Probably the proof that it was a good idea ;-)
>
> IMHO, that concept is *the* ultimate thing to do. A system like Wine
> which allows to transparently run TOS programs on a host OS, without the
> burden of a full-blown hardware emulator. It is the way I would like to
> run old Atari software nowadays (if any). Combined to a JIT CPU
> emulator, it would provide ultimate speed. And full memory protection.
>
> Speaking about RunTOS, this was a very alpha project. Mainly a proof of
> concept. I only implemented a very minimal set of GEMDOS functions,
> partially, but IIRC this was good enough to run LHarc without arguments
> and see its help message. As completing that project was a huge task, I
> never went further. And it was so dirty and experimental that I didn't
> consider to share it. Also, at that time, the legal status of Musashi
> was unclear. This was very long ago, before I got interested in GCC or
> EmuTOS. I even started to write my own 68000 emulator before using Musashi.
>
> TOSEMU (as well as ParaTos) seems to be a good project, as it is written
> from scratch around Musashi. And we see that at least one other people
> has contributed. IMHO, the most important fact is that project has *unit
> tests*. And I see that it already uses Travis CI (it has a .travis.yml).
> So new features can be safely developed and refactored, without the fear
> of breaking something. Definitely, the most sane and modern approach.
>
> Long live to all of those new projects!
>
> --
> Vincent Rivière
> _______________________________________________
> Freemint-list mailing list
> Freemint-list at mail.atariforge.org
> http://mail.atariforge.org/mailman/listinfo/freemint-list
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.atariforge.org/pipermail/freemint-list/attachments/20170124/fc8f3cb4/attachment.html