If an OS was to be written in Python, how'w it look?

[process]

If an OS was to be written in Python and the hardware optimized for
it, what changes would be made to the hardware to accomodate Python
strenghs and weaknesses?

Some tagged architecture like in Lisp machines?
http://en.wikipedia.org/wiki/Tagged_architecture

What else?

[bieffe62]

On 6 Ott, 06:16, process <[e-mail]:

If an OS was to be written in Python and the hardware optimized for
it, what changes would be made to the hardware to accomodate Python
strenghs and weaknesses?

Some tagged architecture like in Lisp machines?http://en.wikipedia.org/wiki/Tagged_architecture

What else?

I would say that a python-processor should have dictionary lookup
(hash tables), garbage collection and dynamic lists implemented by
hardware/firmware.

Maybe in another twenty years ...

Ciao
----
FB

[Eric Wertman]

>If an OS was to be written in Python and the hardware optimized for

>it, what changes would be made to the hardware to accomodate Python
strenghs and weaknesses?

I'm no expert, but this would seem like a good example of something
that python wasn't good for. I have always wondered, though, what a
Linux kernel module would look like that had a python (or java, or
whatever) interpreter running low-level, so the higher level
components on the operating system could be implemented in an
interpreted language. Is there any benefit to something like that?
Or is that crap too? Again, I'm no expert.

Eric

[Dan Upton]

On Mon, Oct 6, 2008 at 11:02 AM, Eric Wertman <ew******@gmail.comwrote:

>>If an OS was to be written in Python and the hardware optimized for
it, what changes would be made to the hardware to accomodate Python
strenghs and weaknesses?

I'm no expert, but this would seem like a good example of something
that python wasn't good for. I have always wondered, though, what a
Linux kernel module would look like that had a python (or java, or
whatever) interpreter running low-level, so the higher level
components on the operating system could be implemented in an
interpreted language. Is there any benefit to something like that?
Or is that crap too? Again, I'm no expert.

The closest I can think of to that is Singularity, Microsoft's
research OS written in .NET (well, C# specifically I guess). I think
their intent was more to shrink the size of the trusted computing base
though and then make all of the actual OS services and such managed
code.

http://research.microsoft.com/os/Singularity/

[Paul Rubin]

"Eric Wertman" <ew******@gmail.comwrites:

I'm no expert, but this would seem like a good example of something
that python wasn't good for. I have always wondered, though, what a
Linux kernel module would look like that had a python (or java, or
whatever) interpreter running low-level, so the higher level
components on the operating system could be implemented in an
interpreted language. Is there any benefit to something like that?
Or is that crap too? Again, I'm no expert.

This was done 20 years ago with the Lisp machine. Really, it was a
bold idea back then, but programming language and systems
understanding have comea long way since then. Best to keep using
Python as a scripting language; it's not the right thing for an OS.

[Paul Rubin]

"Dan Upton" <up***@virginia.eduwrites:

The closest I can think of to that is Singularity, Microsoft's
research OS written in .NET (well, C# specifically I guess).

Singularity is almost the exact opposite of this and I don't think it
uses (unmodified) C#. It does away with use of hardware memory
protection in any form, and relies on rigorous, statically enforced
type safety in the compilers and OS to isolate processes from one
another.

[Gabriel Genellina]

En Mon, 06 Oct 2008 01:16:55 -0300, process <ci**********@gmail.com>
escribió:

If an OS was to be written in Python and the hardware optimized for
it, what changes would be made to the hardware to accomodate Python
strenghs and weaknesses?

There was an experiment ("Unununium"), now abandoned:
http://unununium.org/

--
Gabriel Genellina

[James Mills]

On Tue, Oct 7, 2008 at 4:02 PM, Gabriel Genellina
<ga*******@yahoo.com.arwrote:

There was an experiment ("Unununium"), now abandoned: http://unununium.org/

Yeah does anyone have or know where one
can get the source code and any other
materials relating to Unununium ? It not only
seems to be abandoned, but it's content also lost :/

cheers
James

--
--
-- "Problems are solved by method"

[Gabriel Genellina]

En Tue, 07 Oct 2008 19:30:24 -0300, James Mills
<pr******@shortcircuit.net.auescribió:

On Tue, Oct 7, 2008 at 4:02 PM, Gabriel Genellina
<ga*******@yahoo.com.arwrote:

>There was an experiment ("Unununium"), now abandoned:
http://unununium.org/

Yeah does anyone have or know where one
can get the source code and any other
materials relating to Unununium ? It not only
seems to be abandoned, but it's content also lost :/

web.archive.org contains the site history:
http://web.archive.org/web/*re_/http....unununium.org
Going back to Jan 2007 is enough to discover that their repository was at
http://www.unununium.org/darcs/ - and it's still there.

--
Gabriel Genellina

[James Mills]

On Wed, Oct 8, 2008 at 10:40 AM, Gabriel Genellina
<ga*******@yahoo.com.arwrote:

web.archive.org contains the site history:
http://web.archive.org/web/*re_/http....unununium.org
Going back to Jan 2007 is enough to discover that their repository was at
http://www.unununium.org/darcs/ - and it's still there.

Thanks :)

Is anyone still interested in continuing Uuu's efforts ?
I've just checked out the darcs repository and converted
it to Mercurial. I'm going to have a bit of a play wiht it...

Anyone interested in this ?

cheers
James

--
--
-- "Problems are solved by method"

[Greg Ewing]

James Mills wrote:

I've just checked out the darcs repository and converted
it to Mercurial. I'm going to have a bit of a play wiht it...

Anyone interested in this ?

I'll be interested to hear of any developments.