"Zane H. Healy" <healyzh at aracnet.com> wrote:
On Fri, 5 Jun 2009, Sridhar Ayengar wrote:
> Dave McGuire wrote:
>>>> >>>>> It's damnably slow, but at least it works.
>>> >>>> Yep. The scary thing is...an OS like RSX-11 or RSTS/E is
quite zippy
>>> >>>> on a machine like that.
>> >>>
>> >>> Hmmm. Porting (or writing a work-alike of) RSX-11 for a newer
>> >>> architecture (suck as PC -- for the low cost) comes to mind.
> >>
> >> That would be extremely yummy.
>
> I wonder how hard it would be? I can't say that I'm very familiar with the
> capabilities of RSX, but, considering the size of the code we're talking
> about, could it be possible that it wouldn't involve as much work as writing
> a more modern, featureful, cluttered, bloaty operating system?
The main problem I see with porting it is that to do so would be in
violation of copywrite. The second problem I see is that I'm not sure what
exists in the way of source code. You would need some sort of Macro-11 to
Macro-x86 compilier. An easier port might be something like RSTS/E, though
the source code for it is somewhere around 150MB, IIRC.
RSX ships with full, commented source code. So the OS itself isn't a
problem. Applications are where you'd suffer.
But I don't think a MACRO-11 compiler for x86 would be the solution.
We're talking about the OS here. There is way too much hardware-specific
things going on that needs to be done differently on an x86. I'd rewrite
it, using the same layout of the code and everything, but rewritten to
use the x86 architecture instead. The MMU is different, and lots of code
is playing with that. And registers and register conventions differ, not
to mention processor context, and locking mechanisms.
Realistically time might be better spent on writing a
TCP Stack and running
it under an emulator.
Probably. I already have most of the TCP/IP stack running. :-)
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt at softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol