3D Virtual World Croquet/Cobalt inside VisualWorks

Hi:

Is there interest having the 3D Virtual World framework Croquet/Cobalt running inside VisualWorks? Would there be any major problems making such a port?
http://opencroquet.org

All the best,
Aik-Siong Koh

So far as I know, Croquet uses OpenGL, and there is an active OpenGL interface in VW


James Robertson
Cincom Smalltalk Product Evangelist
http://www.cincomsmalltalk.com/blog/blogView
Talk Small and Carry a Big Class Library




On Dec 7, 2009, at 12:03 PM, askoh wrote:


_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

I'd be interested. Mostly for physics modeling using a computational trick. I have complex interactions to model. For computational efficiency and model simplicity, I'd want to model gravity by varying proportions. The "Turning Gravity Inside Out" video here is a decent beginner level to the underlying theory:

http://www.avantgravity.com/3d_animation.html

I could use squeak for this but I prefer VW.

Paul Baumann


-----Original Message-----
From: [hidden email] [mailto:[hidden email]] On Behalf Of askoh
Sent: Monday, December 07, 2009 12:04 PM
To: [hidden email]
Subject: [vwnc] 3D Virtual World Croquet/Cobalt inside VisualWorks


Hi:

Is there interest having the 3D Virtual World framework Croquet/Cobalt running inside VisualWorks? Would there be any major problems making such a port?
http://opencroquet.org

All the best,
Aik-Siong Koh
--
View this message in context: http://old.nabble.com/3D-Virtual-World-Croquet-Cobalt-inside-VisualWorks-tp26680326p26680326.html
Sent from the VisualWorks mailing list archive at Nabble.com.

_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc


This message may contain confidential information and is intended for specific recipients unless explicitly noted otherwise. If you have reason to believe you are not an intended recipient of this message, please delete it and notify the sender. This message may not represent the opinion of IntercontinentalExchange, Inc. (ICE), its subsidiaries or affiliates, and does not constitute a contract or guarantee. Unencrypted electronic mail is not secure and the recipient of this message is expected to provide safeguards from viruses and pursue alternate means of communication where privacy or a binding message is desired.


_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

An important requirement for Croquet/Cobalt is the replication of computation. Do VisualWorks VMs compute identically on different machines? Same input should give same output for both integer and floating point calculations.

Thanks,
Aik-Siong Koh


<quote author="James Robertson-7">
So far as I know, Croquet uses OpenGL, and there is an active OpenGL interface in VW


James Robertson
Cincom Smalltalk Product Evangelist
http://www.cincomsmalltalk.com/blog/blogView
Talk Small and Carry a Big Class Library

On 2009-12-07 3:42pm, askoh <[hidden email]> wrote:
> An important requirement for Croquet/Cobalt is the replication of computation. Do VisualWorks VMs compute identically on different machines?

You're so wrong. This isn't an "important requirement", it's what Croquet is all about.
Croquet certainly isn't about 3D, which it is bollocks at. A third thing you're wrong about
is thinking anyone gives a shit about replication of computation.

There is no market for it and that's why Croquet is an abject pathetic overhyped failure.
The ultimate total number of people with any kind of interest in replication of computation
(as opposed to data or code) can't be much more than a few dozen programmers on
the entire planet. MMORG engine makers, pretty much.

Croquet is a solution in search of a problem.

Finally, with your gross misconceptions and delusional sense of self-importance
(redirected through Croquet to make it seem less egotistical) you make me sick.
I despise the kind of self-important delusional fanboyism you epitomize.

Croquet isn't just a useless solution in search of a problem, it's a pushy, whiny
little bitch of a solution.

Put me firmly in the Hell No category.
_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

This is a technical mailing list. Please try to keep the discussions civil.

At 06:24 PM 2009-12-07, [hidden email] wrote:

On 2009-12-07 3:42pm, askoh <[hidden email]> wrote:
> An important requirement for Croquet/Cobalt is the replication of computation. Do VisualWorks VMs compute identically on different machines?

You're so wrong. This isn't an "important requirement", it's what Croquet is all about.
Croquet certainly isn't about 3D, which it is bollocks at. A third thing you're wrong about
is thinking anyone gives a shit about replication of computation.

There is no market for it and that's why Croquet is an abject pathetic overhyped failure.
The ultimate total number of people with any kind of interest in replication of computation
(as opposed to data or code) can't be much more than a few dozen programmers on
the entire planet. MMORG engine makers, pretty much.

Croquet is a solution in search of a problem.

Finally, with your gross misconceptions and delusional sense of self-importance
(redirected through Croquet to make it seem less egotistical) you make me sick.
I despise the kind of self-important delusional fanboyism you epitomize.

Croquet isn't just a useless solution in search of a problem, it's a pushy, whiny
little bitch of a solution.

Put me firmly in the Hell No category.
_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

A few weeks ago, I heard a talk from someone from Microsoft Research
about using replication of computation for security.  To make it more
interesting, one version of the computation was done by Javascript,
the other by C#.  I imagine it was fairly non-trivial to get right.

The reason Croquet replicates computation is to decrease the response
time of interactive applications.  It is an interesting use of
replication, though it certainly hasn't caught on yet.

The most common reason for replicating computation is reliability.
This is popular in military systems.  I know several large business
sytems that use it, too.

-Ralph Johnson
_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

Hi guys

if you look for Richard in google you will see that he got banned from a lot of forums for insulting
behavior.
Askoh do not take him personnally. The sad aspect of it is that under his nasty and insulting comments
there are some truth :)

Stef


On Dec 8, 2009, at 12:24 AM, [hidden email] wrote:


_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

Thanks for your meaningful post. Like Smalltalk, replication of computation is not popular, but they have virtues that will go on forever.

Back to my question: Can VisualWorks VMs duplicate computation across different platforms? Can they be made to do so without extreme effort?

All the best,
Aik-Siong Koh

Ralph Johnson wrote

A few weeks ago, I heard a talk from someone from Microsoft Research
about using replication of computation for security.  To make it more
interesting, one version of the computation was done by Javascript,
the other by C#.  I imagine it was fairly non-trivial to get right.

The reason Croquet replicates computation is to decrease the response
time of interactive applications.  It is an interesting use of
replication, though it certainly hasn't caught on yet.

The most common reason for replicating computation is reliability.
This is popular in military systems.  I know several large business
sytems that use it, too.

-Ralph Johnson
_______________________________________________
vwnc mailing list
vwnc@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

For most platforms, they answer is "probably yes".  We do know, however,
of cases in which the standard libraries provided by individual
platforms are a bit at odds with the rest.  For example, on HPUX, the
message floorLog10 may answer incorrect results in a few cases whereas
the rest of our supported platforms do provide correct results.  When we
tracked this down, the root cause was the HPUX compilation environment.
We could fix it by hand so that results are the same across all
platforms.  However, people using the HPUX platform may be confused by
results that do not match the expectations suggested by the platform.

Finally, I'd expect trigonometric functions to be off by a couple bits
or so depending on whether the platform is x86 or SPARC.  In fact, IIRC,
x86/64 implements trigonometric functions differently than x86/32
because, at the time 32 bit x86 FPUs were implemented, a better way to
approximate pi was not known.  This can be a source of different
results.  I do not know, however, that it would necessarily be
appropriate to mimick x86/32 on an x86/64 when the x86/64 results are
better.

Andres.

-----Original Message-----
From: [hidden email] [mailto:[hidden email]] On
Behalf Of askoh
Sent: Tuesday, December 08, 2009 1:26 PM
To: [hidden email]
Subject: Re: [vwnc] 3D Virtual World Croquet/Cobalt inside VisualWorks


Thanks for your meaningful post. Like Smalltalk, replication of
computation is not popular, but they have virtues that will go on
forever.

Back to my question: Can VisualWorks VMs duplicate computation across
different platforms? Can they be made to do so without extreme effort?

All the best,
Aik-Siong Koh


Ralph Johnson wrote:
--
View this message in context:
http://old.nabble.com/3D-Virtual-World-Croquet-Cobalt-inside-VisualWorks
-tp26680326p26701058.html
Sent from the VisualWorks mailing list archive at Nabble.com.

_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

That sounds encouraging. If we just limit ourselves to VMs for Windows, Linux and Mac OSX on the x86 architecture, do they replicate computation already?
Thanks,
Aik-Siong

Valloud, Andres wrote

For most platforms, they answer is "probably yes".  We do know, however,
of cases in which the standard libraries provided by individual
platforms are a bit at odds with the rest.  For example, on HPUX, the
message floorLog10 may answer incorrect results in a few cases whereas
the rest of our supported platforms do provide correct results.  When we
tracked this down, the root cause was the HPUX compilation environment.
We could fix it by hand so that results are the same across all
platforms.  However, people using the HPUX platform may be confused by
results that do not match the expectations suggested by the platform.

Finally, I'd expect trigonometric functions to be off by a couple bits
or so depending on whether the platform is x86 or SPARC.  In fact, IIRC,
x86/64 implements trigonometric functions differently than x86/32
because, at the time 32 bit x86 FPUs were implemented, a better way to
approximate pi was not known.  This can be a source of different
results.  I do not know, however, that it would necessarily be
appropriate to mimick x86/32 on an x86/64 when the x86/64 results are
better.

Andres.

In as much as our floating point tests pass, the answer is yes on all
platforms with at least the following exceptions:

* On HPUX, floorLog10 may not give correct answers for really small
receivers.

* Our trigonometric tests are currently very lax.  I would like to fix
that, but I have not had a chance to do so yet.  I think the results are
identical, or very close to being identical anyway, but I have not
verified this.

* Some other floating point tests also have some degree of tolerance
built into them.  I do not know the underlying actual result accuracy,
or whether the tolerances could be tightened.

Basically, for the essential operations (*, +, - and /) I'd expect
identical results unless something is really weird with the platform in
question.  For transcendental functions however, sometimes the results
do depend on the hardware.  Assuming that our compilation environments
provide reasonable implementations for transcendental functions, I'd
hope to see either identical (or, at worst, near identical) results on
the same FPU family across different operating systems.

Andres.

-----Original Message-----
From: [hidden email] [mailto:[hidden email]] On
Behalf Of askoh
Sent: Tuesday, December 08, 2009 2:09 PM
To: [hidden email]
Subject: Re: [vwnc] 3D Virtual World Croquet/Cobalt inside VisualWorks


That sounds encouraging. If we just limit ourselves to VMs for Windows,
Linux and Mac OSX on the x86 architecture, do they replicate computation
already?
Thanks,
Aik-Siong


Valloud, Andres wrote:
>
> For most platforms, they answer is "probably yes".  We do know,
> however, of cases in which the standard libraries provided by
> individual platforms are a bit at odds with the rest.  For example, on

> HPUX, the message floorLog10 may answer incorrect results in a few
> cases whereas the rest of our supported platforms do provide correct
> results.  When we tracked this down, the root cause was the HPUX
compilation environment.
> We could fix it by hand so that results are the same across all
> platforms.  However, people using the HPUX platform may be confused by

> results that do not match the expectations suggested by the platform.
>
> Finally, I'd expect trigonometric functions to be off by a couple bits

> or so depending on whether the platform is x86 or SPARC.  In fact,
> IIRC,
> x86/64 implements trigonometric functions differently than x86/32
> because, at the time 32 bit x86 FPUs were implemented, a better way to

> approximate pi was not known.  This can be a source of different
> results.  I do not know, however, that it would necessarily be
> appropriate to mimick x86/32 on an x86/64 when the x86/64 results are
> better.
>
> Andres.
>
>

--
View this message in context:
http://old.nabble.com/3D-Virtual-World-Croquet-Cobalt-inside-VisualWorks
-tp26680326p26701702.html
Sent from the VisualWorks mailing list archive at Nabble.com.

_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

I wonder, if the number of primitives is small, one could work with
Doubles, but round off the results at a preset resolution, well inside
the minimum hardware precision, say  by a couple digits. Of course,
floating point issues are always more subtle than one would expect, so
this approach might be entirely misguided.

- Dave

> -----Original Message-----
> From: [hidden email] [mailto:[hidden email]] On
Behalf are
> identical, or very close to being identical anyway, but I have not
> verified this.
>
> * Some other floating point tests also have some degree of tolerance
> built into them.  I do not know the underlying actual result accuracy,
> or whether the tolerances could be tightened.
>
> Basically, for the essential operations (*, +, - and /) I'd expect
> identical results unless something is really weird with the platform
in Windows,
> Linux and Mac OSX on the x86 architecture, do they replicate
computation on
>
> > HPUX, the message floorLog10 may answer incorrect results in a few
> > cases whereas the rest of our supported platforms do provide correct
> > results.  When we tracked this down, the root cause was the HPUX
> compilation environment.
> > We could fix it by hand so that results are the same across all
> > platforms.  However, people using the HPUX platform may be confused
by
>
> > results that do not match the expectations suggested by the
platform.
> >
> > Finally, I'd expect trigonometric functions to be off by a couple
bits
>
> > or so depending on whether the platform is x86 or SPARC.  In fact,
> > IIRC,
> > x86/64 implements trigonometric functions differently than x86/32
> > because, at the time 32 bit x86 FPUs were implemented, a better way
to
>
> > approximate pi was not known.  This can be a source of different
> > results.  I do not know, however, that it would necessarily be
> > appropriate to mimick x86/32 on an x86/64 when the x86/64 results
are
> > better.
> >
> > Andres.
> >
> >
>
> --
> View this message in context:
>
http://old.nabble.com/3D-Virtual-World-Croquet-Cobalt-inside-VisualWorks
_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc

To complete the thought, you basically limit precision for every
calculation at the image level so that every platform has identical
results. And so your cumulative calculations don't drift apart across
platforms.

> -----Original Message-----
> From: Wallen, David
> Sent: Tuesday, December 08, 2009 6:14 PM
> To: [hidden email]
> Subject: RE: [vwnc] 3D Virtual World Croquet/Cobalt inside VisualWorks
>
> I wonder, if the number of primitives is small, one could work with
> Doubles, but round off the results at a preset resolution, well inside
the
> minimum hardware precision, say  by a couple digits. Of course,
floating VisualWorks
> >
> > In as much as our floating point tests pass, the answer is yes on
all
> > platforms with at least the following exceptions:
> >
> > * On HPUX, floorLog10 may not give correct answers for really small
> > receivers.
> >
> > * Our trigonometric tests are currently very lax.  I would like to
fix
> > that, but I have not had a chance to do so yet.  I think the results
are
> > identical, or very close to being identical anyway, but I have not
> > verified this.
> >
> > * Some other floating point tests also have some degree of tolerance
> > built into them.  I do not know the underlying actual result
accuracy,
> > or whether the tolerances could be tightened.
> >
> > Basically, for the essential operations (*, +, - and /) I'd expect
> > identical results unless something is really weird with the platform
in
> > question.  For transcendental functions however, sometimes the
results
> > do depend on the hardware.  Assuming that our compilation
environments
> > provide reasonable implementations for transcendental functions, I'd
> > hope to see either identical (or, at worst, near identical) results
on VisualWorks
> >
> >
> > That sounds encouraging. If we just limit ourselves to VMs for
Windows,
> > Linux and Mac OSX on the x86 architecture, do they replicate
computation example, on
> >
> > > HPUX, the message floorLog10 may answer incorrect results in a few
> > > cases whereas the rest of our supported platforms do provide
correct
> > > results.  When we tracked this down, the root cause was the HPUX
> > compilation environment.
> > > We could fix it by hand so that results are the same across all
> > > platforms.  However, people using the HPUX platform may be
confused by
> >
> > > results that do not match the expectations suggested by the
platform.
> > >
> > > Finally, I'd expect trigonometric functions to be off by a couple
bits
> >
> > > or so depending on whether the platform is x86 or SPARC.  In fact,
> > > IIRC,
> > > x86/64 implements trigonometric functions differently than x86/32
> > > because, at the time 32 bit x86 FPUs were implemented, a better
way to
> >
> > > approximate pi was not known.  This can be a source of different
> > > results.  I do not know, however, that it would necessarily be
> > > appropriate to mimick x86/32 on an x86/64 when the x86/64 results
are
> > > better.
> > >
> > > Andres.
> > >
> > >
> >
> > --
> > View this message in context:
> >
http://old.nabble.com/3D-Virtual-World-Croquet-Cobalt-inside-VisualWorks
_______________________________________________
vwnc mailing list
[hidden email]
http://lists.cs.uiuc.edu/mailman/listinfo/vwnc