Re: [MiNT] FreeMiNT 1.18 release coming up.

[Eero Tamminen <oak@helsinkinet.fi>]

Hi,

On maanantai 04 maaliskuu 2013, Markus Fröschle wrote:
> Don't want to judge if its good or bad, but official Atari docs ("Rainbow
> TOS Release Notes",
> http://dev-docs.atariforge.org/files/Rainbow_TOS_RN_8-7-1989.pdf, pg.
> 24) state its perfectly legal to call the AES from supervisor mode (with
> quite some care, though).
> 
> I do not know what this means for EmuTOS alignment efforts, but I'd
> consider if it's legal for TOS 1.04, it should be legal for EmuTOS as
> well.

That advice doesn't say that it's "legal", I read it more like
"if you do it like this, it may succeed with this specific TOS version".


> I also do not know if there are programs that were actively using
> that "feature" (kludge) but my guess is there were some (assume
> otherwise it wouldn't be documented, probably).

My understanding is that EmuTOS tries to be a clean TOS implementation,
it doesn't implement dirty / deprecated API's unless they're required by
large number of existing programs (i.e. cannot be fixed in the programs
themselves).

EmuTOS for example doesn't implement line-A functions, except for the ones
for turning mouse on and off.


What other programs call appl_init() from supervisor mode?

Is there a test program I could try with all the other TOS versions,
one which does also some extra things after appl_init() so that one
can actually notice if memory got corrupted because appl_init() was
called in supervisor mode?


On maanantai 04 maaliskuu 2013, Markus Fröschle wrote:
> when further reading through it, I actually got the impression that the
> sole purpose of the document was turning bugs to features by just
> documenting them. That's the way it is, however.

EmuTOS definitely doesn't try to be compatible with some specific
TOS versions, and their vaguely documented "features".  Here's
another example of thing that was TOS v4 specific:
	http://sourceforge.net/mailarchive/message.php?msg_id=26835796


	- Eero