Strongtalk VM for Squeak?

I think a project like this would need some champion. Having a
Strongtalk VM would be wonderful but it's likely to be a lot of work and
if I look around I don't see a lot of people with spare time at their
disposal.

Cheers,
   - Andreas

David Griswold wrote:

Andreas Raab wrote:
>
> I think a project like this would need some champion. Having a
> Strongtalk VM would be wonderful but it's likely to be a lot of work
> and if I look around I don't see a lot of people with spare time at
> their disposal.
What if we had a technical lead in the Squeak VM community that could
lead a Sun sponsored group (of 1 or 2).  The work could be delineated in
large chunks - defined in such a way that the work could be done by
several people, at different times. Part-time Squeakers answer tough
questions about Squeak and even code here and there - but most of the
coding is done by Sun. If organized and planned in chunks, maybe it
could be done little by little. More cost would go into the planning of
the project. We'd need to know exactly the status of the current code
and how hard it would be to organize it to start a project.

In this scenario, the Sun team, who understands the code better, can
pass the code knowledge onto the Squeaker(s) that are leading the team.
That way, the Squeak community acquires the knowledge needed to maintain
the code. The squeakers can pass knowledge about the inner workings of
Squeak to Sun.

Granted it would be using the workhorse of this "speculative" Sun team.

The Squeak Foundation could sponsor the "open source" initiative
(website, tracking, server space). What chance is there of getting Sun
to contribute? What can Squeak contribute back to Sun in exchange?
If Sun can't contribute manpower, can they contribute funding to the
Squeak Foundation to put toward experienced coders?

brad

David, you've got quite a number of folks on the list thinking  
seriously on this, right now we're plotting so to speak. Andreas btw  
is *quite busy* with Croquet at the moment, but there are other folks  
here who are much more flexible, so give us a day or so to plot. I'll  
note this list is quite small, and some important folks don't  
subscribe to it, however we're forwarding your note around the  
community right now.



On 15-Sep-06, at 3:31 PM, David Griswold wrote:

--
========================================================================
===
John M. McIntosh <[hidden email]>
Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
========================================================================
===

Hi Dave--

      I would love to see the StrongTalk ideas incorporated into Squeak,
but I'd never want to give up Squeak's wonderful self-hosting property
(the virtual machine is written in Smalltalk, and runnable and
debuggable in Smalltalk). All the C++ associated with the current
StrongTalk makes me wince. :)

      Another current Squeak project, Bryce Kampjes' "Exupery", a
physical-processor code translator, manages to maintain this property.
Could we do this with the StrongTalk ideas? Is there some useful overlap
between StrongTalk and Exupery? (Others please feel free to chime in.)

      This also seems like a good time to mention that I'm looking for
work and that I have extensive experience with the Squeak VM (see, for
example, my "Spoon" project, a minimal Squeak[1]).


      thanks!

-C

[1] http://netjam.org/spoon

--
Craig Latta
http://netjam.org/resume

Damn mail system refuses to send mail to this address.... unless this  
gets through

tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Strange OpCodes: UDF: Use Disk for Frisbee

tim Rowledge wrote:
>
> Damn mail system refuses to send mail to this address.... unless this
> gets through
it did

Hi Craig,

- I wouldn't dream of proposing that Squeak dump its self-hosting
properties.  In the short term, I think a better approach would be to factor
out some kind of common VM interface so that people can plug-in whichever VM
they want.

In the medium term, porting Strongtalk can be done more easily by starting
with just the interpreter and no compiler; the system runs just fine that
way, and then all the really complicated stuff like deoptimization etc can
be deferred.

In the long term, it would be great to rewrite the Strongtalk VM to be
self-hosting and self-debugging.  I had wanted to at least consider this
from the beginning (all that C++ makes me wince too), but I was dissuaded by
the VM guys as it is very much a research topic and we were trying to
develop a commercial product.  The performance of the VM in Strongtalk
(especially the interpreter and compiler) is critical to achieving decent
startup times without noticeable compilation pauses, so at the very least
the host VM must be compiled into some very efficient form.  The Squeak
approach of translating a minimal subset to C might help make that possible.
But it would be a huge project to rewrite the VM in Smalltalk, best done as
research somewhere.

- I haven't looked at Exupery.  I'd have to look to comment.  Performance is
as I have said the critical element, and having a huge chunk of time for
some talented people to do it, which is always the problem.

- As for work, I am working on my own right now, and so have no ability to
hire (and hardly any time to even fiddle with Strongtalk).  If Gilad Bracha,
who is the only one still at Sun, manages to get resources to hire people to
work on this (I wouldn't bet on it given Sun's situation and attitude
towards Smalltalk), that might be a possibility.  BTW, something like Spoon
might be a good candidate for the first port of Squeak to the Strongtalk VM,
if it has a smaller set of demands on the VM.

Dave

m33nw489h2c_h9kpx3nmzf2qe9utb7dajw882bbfhkb9hkt6

Wouldn't you rather have David.Griswold.com as your personal domain?
http://rd.mailshell.com/ad465
Earn up to $3 for each of your friends who signs up with Mailshell!
http://rd.mailshell.com/sp5

... and this for testing if the gmane interface does it again (Lars  
emailed that he cleaned it up).

/Klaus

On Sat, 16 Sep 2006 19:02:24 +0200, tim Rowledge <[hidden email]> wrote:

>  Damn mail system refuses to send mail to this address.... unless this  
> gets through
>
> tim
> --
> tim Rowledge; [hidden email]; http://www.rowledge.org/tim
> Strange OpCodes: UDF: Use Disk for Frisbee
>
>

Hi David,

on Sun, 17 Sep 2006 07:11:33 +0200, you wrote:
I think what Strongtalk VM and Squeak VM already have in common is demand  
for a powerful ObjectMemory implementation (the superclass of Squeak's  
interpreter).

Would you say that this can be a good subproject to start with, all in  
Slang.

/Klaus

Hi,

On 9/17/06, David Griswold
<[hidden email]> wrote:
> [...] for a common interface, we would want something
> that hides all the details of heap structure, object structure, etc, i.e.
> abstracts the VM functionality.

perhaps something like MMTk? It's for Java, though, but it has a
pretty nice abstraction of such structures (and GC algorithms).

Some links:
- the "Oil and Water" paper on MMTk:
http://citeseer.ist.psu.edu/blackburn04oil.html
- MMTk tutorial: http://jikesrvm.sourceforge.net/info/talks/ISMM04-MMTk.pdf

Best,

Michael

[1]