Squeak-dev February 1998

squeak-dev@lists.squeakfoundation.org

110 participants
320 discussions

Re: more prototypes
by jecel＠lsi.usp.br 26 Feb '98

26 Feb '98

I wrote: > With the ability > to drag-n-drop objects, cleaning up the mess you make is > hardly a chore later on. I meant "drag-n-drop slots" which allows you to effortlessly change object structure and inheritance tree. About mirrors, they only handle a small part of the reflection problem, like the classes they replace. They only tell you about object structure, but you also need to be able to "hook into" the message passing mechanism, virtual memory, and so on. BTW, a lot of people don't understand why an object can't handle its own reflection. The problem is that a circle object should "think" it is a circle, and not a bunch of bytes in memory or on disk. That is, its message interface (or protocol) should only include application level things. Adding a system level interface to every single object mixes up things that should be kept separate. So while speciall reflection objects aren't really needed, they are a very good style. -- Jecel P.S. I am sending this from the local university, so it seems that my ISP is the problem with sending to this list.

1 0

Administrivia...
by Christopher Oliver 26 Feb '98

26 Feb '98

Could comeone with sufficient magic change my subscription address from oliver(a)fritz.co.traverse.com to oliver(a)fritz.traverse.com? We are currently renumbering our nets and several customer nets, so things aren't quite where they seem anymore. [Argh!] -- Christopher Oliver Traverse Communications Systems Coordinator 223 Grandview Pkwy, Suite 108 oliver(a)traverse.com Traverse City, Michigan, 49684 (define magic (lambda (f) (lambda (x) x))) (define (more-magic n) (lambda (f) (lambda (x) (f ((n f) x)))))

1 0

Re: Newbie Question Re: Prototypes
by jecel＠lsi.usp.br 26 Feb '98

26 Feb '98

John Maloney wrote: > You can check http://self.sunlabs.com/ but I wasn't able to access > it just now so it may have been taken down. Postscript versions of many > of the Self papers can be found there. There is a mirror at: http://www.cs.ucsb.edu/oocsb/self/index.html A great starting point for prototypes is: http://www.physik3.gwdg.de/~rainer/comp/lang/object-based.html Specially interesting are the links to ultimardrev (an implementation of Self 1.0 in Smalltalk) and prototalk (a kit for creating your own prototype based languages). Unfortunately, I think it wouldn't be trivial to translate these from ParcPlace Smalltalk 4.0 to Squeak. There seems to be a problem with these links, however, so if there is interest I could set up a copy at my site. As for Self, "Its not really dead, Jim... not as long as we remember it." -- Jecel P.S.: for some reason, no mail I send to this list via the my ISP's smpt relay ever makes it (and I get no error messages either), so I am sending this by telnetting directly to c.cs.uiuc.edu (don't try this at home, kids!)

1 0

Re: Newbie Question Re: Prototypes
by Mark Wai 26 Feb '98

26 Feb '98

There is a good article talking about 'Classes versus Prototypes' by Antero Taivalsaari in JOOP Nov/Dec 97 Vol 10, No7. It is a good introductory article that also cover the historical and philosophical background of class based and prototype based approach. I would recommend beginner to read it. -- Mark Wai Frontier Systems Architecture Inc. mailto: mwai(a)ibm.net or:[ mwai(a)frontiersa.com] __

2 1

Re: Newbie Question Re: Prototypes
by Ken Dickey 26 Feb '98

26 Feb '98

>What are prototypes and how would they be used in Squeak? There are a number of object models. Smalltalk has a "class-instance" model, which means that one creates a class (object) which has the shared code (methods) and meta-data (e.g. instance slot information) used to "create instances of the class". You have to create a class before making an instance (object). In a Prototype based OO system, objects are thought of as having both state (instance variables) AND behavior (methods). When you want to create a 'child' of an object you can clone it and then change both its state and behavior without modifying its 'parent'. Depending on the OO model, changes to the 'parent' may or may not show up in the 'child'. In a "pure" Prototype model changes to the 'parent' do NOT show up in the child. In practice one may allow a user to have either behavior as desired. What is this other model good for? It has been used to advantage in multimedia authoring where there are a lot of idiosyncratic objects. When you want to create a singleton object, why should you have to create a separate class? Apple had a long-running multi-media authoring project called 'SK8' (pronounced 'skate') which looked at a lot of user/developers to see how people worked and thought about authoring. They ended up with a very interesting Prototype model. A problem with the current Squeak environment is that when a new user appears, he/she is essentially told "go learn the class libraries and then you can build something". With SK8, one gets a bunch of tool pallets and can do "drag 'n drop" authoring. I.e. grab and plunk down media and then "shift-click" on the visual elements to see/change their property sheet and add behaviors. There is also an "English-like" language which is fairly intuitive. E.g. on MouseDown of me set my fillColor to Green play pingSound set my fillColor to Red tell CharlieRectangle to play myQuickTimeMovie end set the fillColor of every Rectangle on the Stage to mySpecialRenderer See: http://sk8.research.apple.com/sk8/sk8.html Note especially: http://sk8.research.apple.com/sk8/editors.html Unfortunately, SK8 only runs on Macs. I am certainly interested (as I suspect are others) in using Squeak to build a SK8-like development environment and Prototypes would be a big help. Now as soon as Squeak & Morphics get a bit more stable.. Cheers, -Ken

2 1

RE: Squeak as rogue-like gaming platform?!
by Alejandro F. Reimondo 25 Feb '98

25 Feb '98

>Hi! Hi, >Whatever, in these kind of games, there's a virtual world, filled with creatures and items, a hero has to explore and to interact with. It seems quite natural to model this world using an object-oriented paradigm. Yes, but the world is made of Objects it is not object-"oriented". Then is natural to use Smalltalk (an environment or objects). >However, everybody has different preferences and the originator of the ORCS project, Fabio d'Allesi, started with the idea to create a whole new language. Smalltalk is not a language, it is an environment with objects. Then why to create a language if we want to have objects? >The system shall consist of a development environment and a runtime system. In evolutionary systems, is important that evolution never ends... Then always the system can evolve (runtime = development, past < present). > Smalltalk, the syntax can be learn in one day, but Smalltalk, the language and it's environment needs a long time to learn. The major problem IMHO is the fact that you cannot use Smalltalk (even for some very basic tasks) unless you've understood the whole idea behind it. I've been working with Smalltalk, and I have played Role games. Learn the basics to play role games is more difficult to learn that Smalltalk. The number of alternatives for a game like Runequest if very bigger that the number of classes in a commercial Smalltalk. With Smalltalk, we the Smalltalkers have the same problems with people that RolePlayers have. There are people that like role playing, and people that NOT. :-( The percent of people that don't like role playing is equivalent to the % of people that don't like Smalltalk. And I personally think, that the cause is the same. Most of the people are not prepared to work in a virtual world (with no real objects). Most of the people does not aggre to be sub-creators (as undestoood by J.R.R. Tolkien in "Tree and Leaf: including the poem <<Mythopoeia>>") > Basically, my idea is to change the language to give the people something that looks more familiar to them and which still has the semantic of Smalltalk. Provably you must to build more "interesting" objects to role-playing builders. >This is a very important aspect of a language because the syntax is the first thing new people have to interact with. And you always make the product for the customer. And in the case of the ORCS project, the typical customer is the spare-time programmer who'd never manage to create a whole game on its own but who has some very cool ideas, he'd like to add to the existing game. I have done (in 1990) a simulation system with components that must be built with a custon made GUI and behaviour of the component was specified by users (in Smalltalk). The components were connected to build a train locomotive. Each component was programmed and tested in a workbench. The users learn in a VERY short time the mesages they needs to use. They were "ocasional" users and they didn't know about objects. The only thing you must do to guaranty the exit is to build a good framework, not a powerfull language (build power objects, not power languages). Let the language be the language of any object (and evolve with them). >I'd like to reuse all development tools of Squeak without great changes because this would give us a very powerful development environment for FREE. Please, don't work to have less. :-) >The language would need a parcel concept similar to VisualWorks. ASAIK Squeak has still the old-fashioned monolithic image model. Is anybody working on this issue or at least can estimate the work needed to add this to Squeak? Andres Vallaud (sqrmax(a)cvtci.com.ar) and me (alereimondo(a)sugarweb.com) are working on it; but I don't know when we will have something ready to be publicated. >The language would also need namespaces (every modern language should have them, IMHO), but that's easy to add while translating it to Smalltalk. I'd however like to see Smalltalk-based NS, too. Again the question... NameSpaces implemented only as a mechanism to hide objects is simple to implement (see ). But if you think in NamesSpaces as Spaces of diferent instantiations of the VM running with different GC Spaces, needs to reimplement a huge part of the VM. Ale.

1 0

Re: Prototypes
by jecel＠lsi.usp.br 25 Feb '98

25 Feb '98

Steve Dekorte wrote: > Is there any interest in making Squeak prototype-based? Well, I am working on this. -- Jecel

4 3

Re: Prototypes
by David Stes 25 Feb '98

25 Feb '98

On Wed, 25 Feb 1998, Stefan Matthias Aust wrote: > I wouldn't call the result Squeak as it would probably feel differently but > I'd like to see such a system becoming reality. I agree with this.

1 0

Re: Analyst
by Sam Adams 25 Feb '98

25 Feb '98

Erik wrote: << I read in some of the reports from oopsla '97 that Adele Goldberg intended to release Analyst and other old parcplace libraries. When Analyst was released it seemed like an ideal data exploration tool. Does anyone know if this is available yet? >> The Analyst was/is? a product of Xeroc Special Information Systems and not ParcPlace. Its not available yet, nor have had any success in finding a good contact. Any suggestions, Squeakers? Actually, I was the one who expressed an interest in resurrecting the Analyst in a version of Squeak, and Adele offered to help enable this and other "software archeology" by releasing the sources file from Smalltalk-80 version 2.3. After a bit of searching, we have located a copy of the 2.3 sources and are waiting (patiently) for its release. I have a number of class libraries already ported from the KSC libraries that I will release as soon as the portability code from 2.3 is released. These include: - More advanced, fully pluggable SelectionInListViews and CodeViews - PluggableGauges - A DialogBox framework - A very full-featured button framework by Ken Auer - The KSC version of HotDraw for ST80, original code by Ward Cunningham - ObjectWorld, a HotDraw-based greatly simplified clone of Randall Smith's Alternate Reality Kit - A primitive but very useful windowbuilder that generates user understandable code for windows and dialogs - Several custom code generators - Lots of enhancements to base classes - etc. In addition, I am currently working on a port of the DMIX music composition system by Daniel Oppenhiem of IBM Research. Its about 3MB worth of sources. Regards, Sam Sam S. Adams, Distinguished Engineer, IBM Network Computing Software Division tie line 444-0736, outside 919-254-0736, email: ssadams(a)us.ibm.com <<Hebrews 11:6, Proverbs 3:5-6, Romans 1:16-17, I Corinthians 1:10>>

1 0

Analyst
by Erik Gjovik 25 Feb '98

25 Feb '98

I read in some of the reports from oopsla '97 that Adele Goldberg intended to release Analyst and other old parcplace libraries. When Analyst was released it seemed like an ideal data exploration tool. Does anyone know if this is available yet? Erik ><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>< Erik Gjovik Internet: erikgj(a)earthlink.net AOL: ErikGj Go Packers -Twelve Championships and counting ><><><><><><><> '29-'31,'36,'39, '44,'61,'62,'65-'67,'97 <><><><><><><><><><><

1 0

Jump to page: