[self-interest] Paper on Traits.

David Ungar ungar at me.com
Sun Sep 25 18:25:28 UTC 2011


The Self VM work was done jointly at Sun and Stanford. I would be surprised if any Sun patents involved it. The Sun patents were done later, as we were thinking about Java. 

-- David (tapped out on my iPhone; blame it for any typoze;-)

On Sep 25, 2011, at 11:22 AM, Jan-Paul Bultmann <janpaulbultmann at me.com> wrote:

> 
> On Sep 25, 2011, at 19:51 PM, Jecel Assumpcao Jr. wrote:
> 
>>  
>> Jan-Paul,
>> 
>> It is very important to note that some words are used in different ways
>> in different projects. Otherwise you can have some very confusing
>> discussions.
>> 
>> "Traits" in Self come from Adele Goldberg's teaching scheme for early
>> Smalltalk where she purposely used different words than the standard
>> jargon to informally introduce ideas before moving on to the actual
>> details for the language. So she would first get people to think about
>> which "traits" an object might have before showing them the "class" that
>> implemented the idea.
>> 
>> Beyond "traits", Self also has "mixins". These names are just
>> conventions and at the language level we just have parent objects, but
>> the different names indicate the different intent behind these objects.
>> 
>> Several languages, including Strongtalk which evolved from Self, have
>> "mixins" as an important feature. So when the research group at Bern
>> decided to add to Squeak a modified version of "mixins", they decided to
>> use a different name so their changes wouldn't be overlooked. That is
>> what "traits" in Squeak or Pharo are. This has been copied in several
>> other projects and languages and so is currently what is normally
>> understood by the term. This is, however, different from what is
>> described in the paper that David indicated.
>> 
>> Back when JIT implementations were less common, one term that caused a
>> lot of confusion was "compiler". For Smalltalk-80 the term meant the
>> source the bytecode translation program while for Self that was called
>> the "parser" and "compiler" meant the bytecode to native machine code
>> translator instead.
>> 
>> > Btw, Davidmay I humbly ask you on your thoughts on Io as a language ^^?
>> > Another grab bag one, or maybe a worthy successor to Self if enhanced
>> > with its Environment ;)?
>> 
>> I am not David (either one ;-) , but I have been a part of the Io
>> community since the very beginning. This is another case of the same
>> words being used differently and causing confusion. Io comes from the
>> Newtonscript tradition of "prototypes", not the Self style. Even Steve
>> Dekorte was confused about this for a very long time, and I can't blame
>> him since when I first read "The Power of Simplicity" paper I also
>> imagined something like Io rather than what Self actually is. But those
>> ideas are actually from Henry Lieberman's 1986 "Using Prototypical
>> Objects to Implement Shared Behavior in Object-Oriented Systems" paper
>> instead.
>> 
>> 
> Yeah the naming clash is amazing especially when one works with projects that used terms before everybody else did^^
> Thanks for the insight on the different origins :D
> 
>> While I like Io, it has never had the kind of implementation technology
>> that Self has always had and which recently has also come to Javascript.
>> 
> Yeah we're kinda working on that with the recent acute implementation^^
> One problem with this is though that Oracle still holds a lot of Davids original patents so were a bit hesitant about implementing all the power of Self.
> Most of them run out in 2014 I think, there are a lot of projects though that are bold enough to simply implement them, but with Oracles recent suing happines its kinda difficult ^^
> 
>> That community is actually smaller and currently less active than the
>> Self one, so I wouldn't expect world domination from that direction any
>> time soon.
>> 
> Smaller yes, less active no. :) You should really visit the #io channel on irc.freenode.net
> 
> Io has the advantage that it is conceptually simpler than Self, so there are more concepts that can arise out of a small base language, like resends and annotations.
> There are vastly parts missing though like Morphic and the Image. But it has a far superior concurrency model.
> Interestingly we can get a Self like look with acute ^^ by encouraging small number of arguments and having a single arg syntax.
> list(1,2,3) at:1 data
> list(1,2,3) at:1 put:5
> As at: returns a listIndex.
> 
> Currently we are working on a VM for Prototype languages, it could be interesting to combine the small working force of the Self and Io communities.
> When lookup is just a slot on Objects (as we plan to do it) it would be trivial to implement different Inheritance models and thus support both languages :) the rest is just syntax^^
> 
> Cheers Jan
>> 
>> -- Jecel
>> 
> 
> 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.selflanguage.org/pipermail/self-interest/attachments/20110925/52d5e7b1/attachment.html>


More information about the Self-interest mailing list