What is Squeak? (Was Re: A roadmap for 3.9)

stéphane ducasse ducasse at iam.unibe.ch
Mon Dec 13 07:57:33 UTC 2004


Hi jos

I agree 200%


>> I think that Marcus is dead-on with this comment.  People shouldn't 
>> be surprised if parts of the system that they contribute get 
>> abandoned if they're not there actively promoting and maintaining 
>> them.  (And I disagree with Andres:  Andreas is actively promoting 
>> and maintaining his pieces of Squeak, and as an active developer, 
>> it's certainly part of that role to critique what others do and 
>> encourage what he wants to see.)
>>
>> But I think that Lex's point is also quite valid: What about the 
>> newcomers to Squeak?  What is their expectation about Squeak, and do 
>> we them (and the community) a disservice by not making some effort to 
>> meet that expectation?  Expectation failure doesn't encourage people 
>> to join a community.  The important question for the community, then, 
>> is to define: What is Squeak?  By answering that, we can more 
>> effectively promote that definition and encourage the appropriate 
>> expectation.
>
> You are speaking of the expectation failure of someone who used 3.6 in 
> class, and later downloads 3.7 and finds that Wonderland doesn't work? 
>  This is a valid point; this we should strive to avoid this.
>
> It's unfortunate that this bug slipped through, especially since it 
> was an easy one to forsee and prevent.  However, 3.7 had a long 
> release cycle, and there was plenty of time for Wonderland users to 
> notice this extremely obvious bug before the final version was 
> released.  Perhaps the SCG folks should have caught this error, but 
> since they didn't, someone who cares about Wonderland should have.
>
>> I'm biased here, but I think that one of the ways that people 
>> discover Squeak is through the OOPSLA paper by Dan et al. and the 
>> White and NuBlue books.  We certainly don't want to let EVERYTHING in 
>> those publications define Squeak -- that would completely limit the 
>> community's ability to change.  But I do think that the NuBlue book's 
>> title, "Squeak: Open Personal Computing and Multimedia" is a pretty 
>> good definition, and one that the other publications agree with.  
>> Squeak is about open personal computing and multimedia.
>>
>> That's what concerns me about the current process in Squeak -- it's 
>> setting aside the personal computing and multimedia aspects (for now 
>> -- I do understand that) in favor of improving the underlying base.  
>> I understand that current members of the community consider those 
>> "goodies" (such as Wonderland and eToys) to be "hacks," but those 
>> "hacks" brought in many people to Squeak.
>
> Very true.
>
>> I do appreciate what Stef and the Berne group have brought to Squeak, 
>> and I think that the environment that they propose for 3.9 sounds 
>> like an exciting one to work in.  But here's my suggestion: It's not 
>> Squeak, at least not as it has been defined and communicated in the 
>> past.  When the base is improved and the personal computing & 
>> multimedia "goodies" are ported back (if they are), then it might be 
>> Squeak again.  But as Marcus points out, that will only happen of the 
>> multimedia developers are still around then, and they might not be 
>> during the interim -- it's not clear that people interested mostly at 
>> the level of the base image are the same kind of people who want to 
>> build things like eToys and Wonderland.
>
> You seem to be equating people who are primarily concerned with 
> improving the system kernel to people who are interested in developing 
> on an improved base system.  The former group is probably not 
> interested in building multimedia apps.  However, many multimedia 
> developers (myself included) fall in the latter group.  There have 
> been many times I've wished for Traits, and had to use a less elegant 
> design to fit with in the constraints of a single-inheritance 
> hierarchy.  And what multimedia developer wouldn't want Squeak running 
> on a JIT compiler?
>
>> I make two concrete proposals -- they're alternatives:
>> A. Call the new thing something else.  Let "Squeak" end at Version 
>> 3.7 or 3.8, unless someone wants to continue it as a tool for 
>> personal computing and multimedia.  Don't let the expectations of 
>> "Squeak" limit where the current community wants to go.  Use the new 
>> name to attract new attention (maybe get Slashdot to notice?) and to 
>> signify a new set of emphases.
>> B. Or, call the 3.9 version "Squeak 4.0," and make it clear that 
>> there is no promise of compatibility or multimedia features across 
>> the boundary from 3.X->4.0.  Say that clearly on the Website, and 
>> make the final 3.x version forever available.  If people want 
>> "personal computing and multimedia," they can download the final 3.x. 
>>  If they want the coolest open source Smalltalk on the planet, with 
>> the base hooks to grow one's own personal computing and multimedia 
>> (like the really interesting eToy/Wonderland substitute whose URL 
>> Marcus sent around), then let them grab the latest 4.x version.
>
> I understand where you're coming from, but this seems a bit of an 
> extreme reaction to a bug in Wonderland that required a 1-line fix (or 
> are there other multimedia features of Squeak that are also broken?).
>
> Squeak has the potential to reach many millions, or even billions of 
> people, but not if gratuitously forked.  If there were deep 
> architectural changes that made it difficult to "port" the multimedia 
> apps forward, then I might be in favor of either one of your 
> proposals.  However, it seems that the required fixes are very small 
> and easy.
>
>> If a day comes when the "goodies" get folded back in, maybe we can 
>> re-merge.  But nobody should hold their breath waiting for it.  The 
>> Georgia Tech group and Andreas' Croquet group can decide which 
>> version(s) they want to develop from, and perhaps fork if they want.  
>> (FYI, the "Scratch" project at the MIT Media Lab is building on 
>> Squeak 2.7 -- the forks are already happening, so we might as well be 
>> honest about it and stop battling over the name.)
>
> Noting that forks are already happening is not an argument for further 
> forks.  As Stephane has repeatedly said, it would have been less work 
> to fork off his own version to support his research, but he instead 
> wanted to allow the broader community to be able to use the SCG's 
> work.  Forking is the easier path when you are pursuing a research 
> agenda.  It takes more effort to integrate the best of everyone's work 
> into a single system, but the synergistic effects are potentially far 
> greater.
>
>> But by making a clear break with the past, Stef and the Berne group 
>> have a freehand to take the base image in the directions that they 
>> want, and people who come to Squeak with the "personal computing and 
>> multimedia" expectation can make a choice.
>
> Isn't that what the Squeakland image is for?  That seems to be where 
> much of the development in that area goes on.
>
> Joshua
>
>
>> Mark
>> __________
>> Mark Guzdial : Georgia Tech : College of Computing/GVU
>> Atlanta, GA 30332-0280
>> Collaborative Software Lab, http://coweb.cc.gatech.edu/csl
>> http://www.cc.gatech.edu/~mark.guzdial/
>>
>>
>
>




More information about the Squeak-dev mailing list