[net.micro.atari16] OS9 mudules to run in GEM

SOFPJF@UOGUELPH.BITNET (Peter Jaspers-Fayer) (09/30/86)

Eric, you don't seem to realize:  OS9 'IS AN OPERATING SYSTEM'.  It's
modules are all re-locatable, use OS9 I/O calls, and in short do not
'fit' onto GEM - NORMALLY.  But I seem to recall (when I used my trusty
old 6809 a lot more) that the assembler (and at least one C) could gen
'Native' code.  But this made testing a pain, as you had to re-boot the
old OS, run your code (if you were lucky you had a debugger), and then
re-boot OS9 to re-EDIT, COMPILE, LINK... sad.  In short, you stayed in OS9.
But then the 'native' OS on the Tandy Color Computer with 64K (there I've
admitted it!) was BASIC (yech), so it wasn't worth it.

   In short: When you boot OS9, you can still run BIOS, but you kiss TOS,
and GEM goodbye (I think).  And as you say, I doubt OS9 has the libraries.
Now, weather the ST version of OS9 has a SIMILAR type of windowing and
menu system... I don't know, havn't seen the ST version, and I don't know
how far it has gotten in it's evolution.  Comments anyone?
 /PJ

olson@endor.harvard.edu (Eric Olson) (09/30/86)

I realize that OS/9 is an operating system unto itself.  The point is that
GEM and VDI are not:  they are acually libraries.  It's still not clear
to me whether they are in ROM on the ST or linked in with every application,
but, regardless, OS/9 should be able to talk to these libraries.  Theoretically
code linked to access GEM/VDI (if such libraries existed) would still run
if the program were run under TOS.  The I/O, on the other hand, is more
troublesome.  I wasn't totally serious in asking such a thing, just wishful
thinking.

-Eric

braner@batcomputer.TN.CORNELL.EDU (braner) (10/01/86)

[]

I don't see why an ST implementation of OS9 couldn't include access
to the GEM graphics routines (VDI, and perhaps parts of AES).  These
are in ROM anyway, might as well use them, or allow OS9 programs to
use them.  (Portability?  Ever tried running a graphics program on
an 'IBM-compatible' just to find out you've got the wrong graphics
card?  'Compatibility' is a myth!)

- Moshe Braner

rb@cci632.UUCP (Rex Ballard) (10/03/86)

In article <1138@batcomputer.TN.CORNELL.EDU> braner@batcomputer.UUCP (braner) writes:
>[]
>
>I don't see why an ST implementation of OS9 couldn't include access
>to the GEM graphics routines (VDI, and perhaps parts of AES).  These
>are in ROM anyway, might as well use them, or allow OS9 programs to
>use them.  (Portability?  Ever tried running a graphics program on
>an 'IBM-compatible' just to find out you've got the wrong graphics
>card?  'Compatibility' is a myth!)
>- Moshe Braner

Actually, it's even easier than it looks.  Both GEM and OS-9 use
"vector tables" to access their "BIOS" routines.  OS-9 could
simply be an additional group of vector tables.  OS-9 also has
a nice capability in that it can use traps to access it's own,
or somone elses routines.  The basic ingredients of VDI are
not that different, perhaps a few arguments would have to be
moved from one place to another.

As to 'IBM-compatible', anyone who has ever made a "plug compatible"
accessory for IBM knows how risky a proposition this can be.  I've
worked for two different companies that stopped working for IBM
when the risks didn't pay off.

The main advantage of OS-9 is that object code compatibility can
be obtained without even so much as re-compilation.

Speed of execution might be a minor issue, but probably not significant
with processors running nearly 1 mips, and actual bit-bumping being
done by other chips.