Re: Ada Operating System

["Marin David Condic" <dont.bother.mcondic.auntie.spam@[acm.org>]

Here's another way for an Ada OS to distinguish itself while also making it
Ada-ish - encouraging development for it in Ada. If a spec was provided that
described a protected type for communicating between processes, wouldn't
that make for something kind of unique and easily connected-to only by Ada
programs? And if it was based on/associated with an Ada spec for generic or
tagged record based messages then again, you'd have something different that
was a natural encouragement to use Ada. Plus with a lot of built-in
intra/inter-process message passing capabilities, the OS would be naturally
suited to OOProgramming and networking.

It doesn't necessarily have to be big or address every possible need. It
only has to be different and useful.

MDC
--
Marin David Condic
Senior Software Engineer
Pace Micro Technology Americas    www.pacemicro.com
Enabling the digital revolution
e-Mail:    marin.condic@pacemicro.com


"Darren New" <dnew@san.rr.com> wrote in message
news:3C9F5DBE.4B71408F@san.rr.com...
> Actually, I think there's another way that an Ada-based OS could be
> superior to a C-based OS. Imagine an OS where all IPC is done via
> message passing of messages between queues. That is, when you (for
> example) want to read a file, you put the file handle and the "read"
> token at the start of a buffer, pass the buffer to the queue for the
> file system thread. The file system thread wakes up, fills up your
> buffer, and passes it back to you. In a protected-memory system, it
> would be helpful if said buffers were always allocated on page
> boundaries, with at most one buffer per page, so simple memory-map
> manipulations could move the buffer from one address space to the other.
>
> Would an Ada compiler (perhaps aided and abetted by the linker) be able
> to take objects of the "Message" generic (or tagged) type and assure
> they're always allocated on page boundaries, and objects of the "Queue"
> generic (or tagged) type are always in pages together (so the OS knows
> where they are perhaps) in a way that would be tedious and error-prone
> in C? Could it be done with appropriate storage pools, so it doesn't
> even really need to be specialized in the compiler?
>
> --
> Darren New
> San Diego, CA, USA (PST). Cryptokeys on demand.
>       Remember, drive defensively if you drink.