Project Direction

[David Griswold]

I'm sure some of you have noticed that I've been mostly off the list for the last few months- my apologies. 

I became fairly discouraged after it became apparent to me late last year that there wasn't a critical mass of people with the right kind of experience ready to pick up the Strongtalk system and run with it, at least in terms of developing and supporting the system as a whole right now.  Not that I ever thought there was that high a probability of that happening immediately, but it was worth a shot to see if the Smalltalk community would rise to the occasion.  Dan Ingalls even offered a $5,000 reward last year for someone to port Squeak to the Strongtalk virtual machine, but there were no takers. While the Smalltalk community as a whole did not rise to the occasion, those of you on this list are the exceptions to that, and I thank you for your interest, enthusiasm, and participation so far.

 

You have to understand that no one from the original Strongtalk project, me included, has the time or desire to pick up that load.  In the years when Strongtalk sat on a shelf at Sun, we all moved on to other, non-Smalltalk-related things.  I got involved again because the overall vision for the system was originally mine, so I probably care the most about it, and I wanted to make sure that the design philosophy was carried on, and that someone was available to answer questions. But this is still very much a side project for me, although I absolutely still believe to this day, that there is still no programming language and virtual machine as advanced as Strongtalk.

 

The big problem at the moment is that the system is far from complete, and there are large, highly complex parts of the system that there is no one to work on, for example

• The user interface needs major work to update it, finish it, and port it.  I could do this, but I simply don't have the time.
• The typechecker needs significant design changes that are probably not going to happen soon.
• A debugger needs to be written, and I don't have the time to do this, either.

The bottom line is that in the near term I don't think we are going to be able to really support Strongtalk as a complete system that is usable by developers.  I considered giving up my involvement in the project, but after a lot of thought, I've concluded that there is still a path forward.

 

Development Direction

• So how can we get Strongtalk on the path to being a supportable, complete system, or at the very least, something that survives and is of use to people, not just as a demonstration?  For now, I think we need to defer any attempt to support the entire system right now, and focus almost entirely on the following three things:
• Increasing VM reliability
• Porting the VM to UNIX-like operating systems like Linux and MacOS
• Porting Squeak to the Strongtalk VM

For Strongtalk to survive, we need a larger community to become enthusiastic about it, and convinced of the need for it to survive.  I think the only feasible way of accomplishing that right now is to make the Strongtalk virtual machine a usable alternative to the Squeak virtual machine.  To do that it will need to be grafted to Squeak, made more reliable, and made to run on some major platforms that Squeak is used on.

 

Once a larger group of people see how fast the Strongtalk VM actually is, running a code base that they are familiar with, I believe we would create a lot more interest in the Strongtalk technology.  Also, in the short term, we would get a finished, supported Smalltalk stack sitting on top of the VM.

 

Hopefully, this would help get a critical mass of developers interested in working on the virtual machine.  Once that happens, many other things would become easier, including spending more time on development of the Strongtalk code base above the virtual machine.  But for that, time will tell.

 

At the moment, it looks like there is starting to be a reasonable group of you folks out there working on how to port the VM, so that's a good start- thanks again.  I'll start trying to consolidate your changes into the trunk where that is appropriate.  Otherwise, I'm going to start focusing on a Squeak port.  It looks like we have the possibility of getting a little money to pay someone to help out with that from Summertalk, so hopefully we can get going on that pretty quickly.

Cheers,

-Dave

[stephane ducasse]

On 1 mai 07, at 02:22, David Griswold wrote:

> I'm sure some of you have noticed that I've been mostly off the
> list for the last few months- my apologies.
> I became fairly discouraged after it became apparent to me late
> last year that there wasn't a critical mass of people with the
> right kind of experience ready to pick up the Strongtalk system and
> run with it, at least in terms of developing and supporting the
> system as a whole right now. Not that I ever thought there was
> that high a probability of that happening immediately, but it was
> worth a shot to see if the Smalltalk community would rise to the
> occasion. Dan Ingalls even offered a $5,000 reward last year for
> someone to port Squeak to the Strongtalk virtual machine, but there
> were no takers. While the Smalltalk community as a whole did not
> rise to the occasion, those of you on this list are the exceptions
> to that, and I thank you for your interest, enthusiasm, and
> participation so far.

Hi david

in fact this was not clear to me that dan was offering that. Because
if this is still correct we can
really advertise it. ESUG could even add some money for that.
Could you contact dan and really ask him?

[Marten]

On 1 Mai, 02:22, "David Griswold" <David.Grisw...@acm.org> wrote:

> that I ever thought there was that high a probability of that happening
> immediately, but it was worth a shot to see if the Smalltalk community would
> rise to the occasion. Dan Ingalls even offered a $5,000 reward last year

I think, that the commercial Smalltalk vendors have other problems and
therefore
there is little interest in looking at the VM technology or stuff like
this.

I've posted about Strongtalk in the VASmalltalk forum, but the result
were
more or less pretty clear: Smalltalk speed is not criticial for lots
of
customers any more - its nice to have it faster, but its not critical
any
more - only a few persons making heavy computations would love to see
some
more speed.

For the other thing: the optional type system: well I would love to
see it in
VASmalltalk, but I assume, that type systems are not a hot topic at
all
in the community these days.

Therefore the community has other things to do ...

Marten

[dave....@gmail.com]

I think one of the problems (ok, actually, my biggest problem) is not
knowing who is working on what. Does the google wiki support a task
list of any sort of project management feature that we could use?

I can volunteer 3-4 hours a week to work on some aspect of the items
you have identified. I'll pick up my work once the various changes
that have been contributed by prunedtree, Thiago and others have been
merged back to Trunk -- heck, I can even help with sorting through
those if you'd like.

Just let me know.

-- the other dave.

> a.. The user interface needs major work to update it, finish it, and port

> it. I could do this, but I simply don't have the time.

> b.. The typechecker needs significant design changes that are probably not
> going to happen soon.
> c.. A debugger needs to be written, and I don't have the time to do this,

> either.
> The bottom line is that in the near term I don't think we are going to be
> able to really support Strongtalk as a complete system that is usable by
> developers. I considered giving up my involvement in the project, but after
> a lot of thought, I've concluded that there is still a path forward.
>
> Development Direction

> a.. So how can we get Strongtalk on the path to being a supportable,

> complete system, or at the very least, something that survives and is of use
> to people, not just as a demonstration? For now, I think we need to defer
> any attempt to support the entire system right now, and focus almost
> entirely on the following three things:

> a.. Increasing VM reliability
> b.. Porting the VM to UNIX-like operating systems like Linux and MacOS
> c.. Porting Squeak to the Strongtalk VM

[David Griswold]

Hi Stef,

> -----Original Message-----
> From: strongtal...@googlegroups.com
> ... Dan Ingalls even offered a $5,000 reward last year for
> > someone to port Squeak to the Strongtalk virtual machine, but there
> > were no takers. While the Smalltalk community as a whole did not
> > rise to the occasion, those of you on this list are the exceptions
> > to that, and I thank you for your interest, enthusiasm, and
> > participation so far.
>
> Hi david
>
> in fact this was not clear to me that dan was offering that. Because
> if this is still correct we can
> really advertise it. ESUG could even add some money for that.
> Could you contact dan and really ask him?

Unfortunately, Dan's offer was only good until the end of 2006. I think
he wanted to get some momentum going, but the response was so
disappointing that I would feel uncomfortable asking him, since the
offer was such a generous one to begin with.

-Dave

[David Griswold]

Hi Dave,

> -----Original Message-----
> From: dave....@gmail.com
>
> I think one of the problems (ok, actually, my biggest problem) is not
> knowing who is working on what. Does the google wiki support a task
> list of any sort of project management feature that we could use?

Yes, we need to get more organized. Google Code doesn't really have
anything that really addresses that, but here's a suggestion: for each task
that someone is working on, there should be (and is, I hope) an item in the
issue database. When someone starts working on that, I suggest they add a
comment to that issue saying that they are working on it, so that others
know. Also, when code is submitted, it should be added as a comment with an
attachment, to the issue with which it is associated. This would be better
than submitting code as e-mail to the development list, and would give us a
clear trail of the submissions for each issue. If everyone likes that idea,
I'll add it to the wiki as part of our development standards.

> I can volunteer 3-4 hours a week to work on some aspect of the items
> you have identified. I'll pick up my work once the various changes
> that have been contributed by prunedtree, Thiago and others have been
> merged back to Trunk -- heck, I can even help with sorting through
> those if you'd like.

In fact, the mess I'm going to try to sort out next, is catching up and
consolidating all the various changes that have been floating on the list
over the last several months, so that I can review and commit them. If you
want to help with figuring out a consolidated set of changes and who made
them, that would be fantastic. I'll review and commit them ASAP. Of course
this is my fault for being off the list :-(

Cheers,
- Dave