Alan Kay's new project

[Mark Haniford]

I'm wondering what people's thoughts are regarding Alan Kay's new project.  http://www.stanford.edu/class/ee380/Abstracts/070214.html

I just downloaded Pepsi, which is interesting in that it somewhat like Objective-C in that you can mix C with Smalltalk-like code.  There's another language in the works called Coke (know less about), but seems to have a Lisp-like syntax.  But I guess the real interesting thing about Pepsi, Coke, ALBERT is that it seems its goals are to have easy different object-model integration.

It looks like Kay got 5 million from the NSF and another 5 million in private funding for a 5 year project.  I wonder if Slate could be worked into this system.

[Brian Rice]

On Feb 8, 2007, at 8:37 PM, Mark Haniford wrote:

> I'm wondering what people's thoughts are regarding Alan Kay's new  
> project.  http://www.stanford.edu/class/ee380/Abstracts/070214.html

I'm aware of it. I wouldn't call it Alan Kay's project. It's Ian  
Piumarta's, and he's lucky enough to be one of the people that Alan  
Kay will listen to.

> I just downloaded Pepsi, which is interesting in that it somewhat  
> like Objective-C in that you can mix C with Smalltalk-like code.  
> There's another language in the works called Coke (know less  
> about), but seems to have a Lisp-like syntax.  But I guess the real  
> interesting thing about Pepsi, Coke, ALBERT is that it seems its  
> goals are to have easy different object-model integration.
>
> It looks like Kay got 5 million from the NSF and another 5 million  
> in private funding for a 5 year project.  I wonder if Slate could  
> be worked into this system.

Eh, perhaps. The whole thing looks very quirky and Ian-centric to me,  
and he has a long history of not finishing what he starts. I'm still  
in wait-and-see mode, personally. Also, don't forget that research  
projects have absolutely no commitment to be useful or forthcoming to  
other open-source hackers, so don't take it for granted that we can  
just join in on the fun - after all, Squeak was "open-source" for 6  
years before there was a community not directly attached to Alan that  
had *any* say over its direction.

I'd prefer a situation where we can take the lessons from there and  
morph Slate along those lines (and hopefully even re-use some code),  
and as I've stated before, I don't care how much Slate has to change  
if it means that the idea succeeds.

Coke/Pepsi also seem very single-dispatch-centric, for the record.

I invite comments/arguments/suggestions as usual.

--
-Brian
http://briantrice.com

[Mark Haniford]

On 2/11/07, Brian Rice <water@...> wrote:

On Feb 8, 2007, at 8:37 PM, Mark Haniford wrote:

> I'm wondering what people's thoughts are regarding Alan Kay's new
> project.   http://www.stanford.edu/class/ee380/Abstracts/070214.html

I'm aware of it. I wouldn't call it Alan Kay's project. It's Ian
Piumarta's, and he's lucky enough to be one of the people that Alan
Kay will listen to.

Alan Kay is the name recognition behind the proposal that got the NSF funding.  But yeah, ALBERT wouldn't be Alan Kay's baby.
 

Eh, perhaps. The whole thing looks very quirky and Ian-centric to me,
and he has a long history of not finishing what he starts.

 I don't know what quirky and Ian-centric means, but Slate never got finished either...
But at least they have funding, which means there's a high likelyhood that something will be produced, unlike some sourceforge crap.

I'm still
in wait-and-see mode, personally. Also, don't forget that research
projects have absolutely no commitment to be useful or forthcoming to
other open-source hackers, so don't take it for granted that we can
just join in on the fun

It was my understanding that the system would be somewhat language agnostic.  Now whether it is useful for Slate or not is another matter, but I don't know why you would think that anybody thinks that research projects have some kind of obligation to open source hackers.
 

- after all, Squeak was "open-source" for 6
years before there was a community not directly attached to Alan that
had *any* say over its direction.

If it's open source then anybody has much say  over the code as they want.  I'm not sure what you're getting  at, besides maybe some  grudge against Alan Kay.

I'd prefer a situation where we can take the lessons from there and
morph Slate along those lines (and hopefully even re-use some code),
and as I've stated before, I don't care how much Slate has to change
if it means that the idea succeeds.

Agreed, and that's why I posted, so that any code that can be leveraged in order to implement the ideas of Slate can be looked at.
 

Coke/Pepsi also seem very single-dispatch-centric, for the record.

Yeah I know, it's limiting.  I'm off to hack some Dylan code.

--
-Brian
http://briantrice.com

[Brian Rice]

On Feb 11, 2007, at 11:47 AM, Mark Haniford wrote:

>> I'm still
>> in wait-and-see mode, personally. Also, don't forget that research
>> projects have absolutely no commitment to be useful or forthcoming to
>> other open-source hackers, so don't take it for granted that we can
>> just join in on the fun
>
> It was my understanding that the system would be somewhat language  
> agnostic.  Now whether it is useful for Slate or not is another  
> matter, but I don't know why you would think that anybody thinks  
> that research projects have some kind of obligation to open source  
> hackers.

Don't set me up as a straw man - I don't believe that others *think*  
that research projects have such an obligation, but the mere idea  
that this would be useful to anyone but those who use their artifact  
(the exact piece of code as they release it) means that there is a  
social connection which may not be satisfactory because of different  
implicit expectations.

>>  - after all, Squeak was "open-source" for 6
>> years before there was a community not directly attached to Alan that
>> had *any* say over its direction.
>
> If it's open source then anybody has much say  over the code as  
> they want.  I'm not sure what you're getting  at, besides maybe  
> some  grudge against Alan Kay.

The problem for years was that Kay's team moved in whatever direction  
they liked without reference to those who were using Squeak outside  
of Disney, except to product versioned releases. Sure, we could hack  
on Squeak, but the next revision might make everything bizarrely  
incompatible or combine compelling additions with algorithmic slow-
downs and complexities in Morphic that were not documented for a  
while and which the core team kept resisting suggestions or patches for.

I don't have a grudge, but it bears repeating that a license is not  
sufficient for a satisfying community experience.

For example, there are cleanups to Squeak's I/O system from 1997 and  
maintained as a fork for all these years that I adopted in Slate  
easily but still are not main-line Squeak features. I also spent a  
lot of time trying to divine Self's secrets and not getting far  
because the released source wasn't clear or concise enough at all.

I guess I should just simply state that there are disadvantages to  
the publicity addition of Alan Kay, in that it draws a large and  
creative crowd that then doesn't necessarily find a workable social  
situation ("too many cooks spoil the broth" combined with multiple  
statements by Kay himself that he doesn't care that much about  
Smalltalk or Squeak as platforms).

>> I'd prefer a situation where we can take the lessons from there and
>> morph Slate along those lines (and hopefully even re-use some code),
>> and as I've stated before, I don't care how much Slate has to change
>> if it means that the idea succeeds.
>
> Agreed, and that's why I posted, so that any code that can be  
> leveraged in order to implement the ideas of Slate can be looked at.

Cool, well I'm keeping my eye open, and hopefully we'll be able to  
see some of the purported advancements and make use of them easily. I  
really want this to work, I just have to state my doubts... a lot of  
my experiences in the last few years have re-iterated for me that the  
most important thing about any project (commercial or open-source) is  
the quality of the people and how socially gracious they are, not  
just in their words but their actions.

Ian has dropped the ball before on JIT compilers and VM projects, and  
Slate also has obviously hit a huge snag with Lee's drop-off in  
enthusiasm even though I have faith in the worthwhileness of the idea.

And on the other hand, maybe I'll just figure out how to layer PMD +  
Slate-ish syntax on that run-time and be happy with that.

>> Coke/Pepsi also seem very single-dispatch-centric, for the record.
>
> Yeah I know, it's limiting.  I'm off to hack some Dylan code.

:)

--
-Brian
http://briantrice.com