Metaclasses and co.
Locked
 
|
Today I spend some time to read and understand an old mail from
Andreas (http://lists.squeakfoundation.org/pipermail/squeak-dev/2005- October/096370.html) about having a parallel metaclass hierarchy. I really like this. Thanks Andreas!! I wrote a short test to understand it, I think it might interest some of you: -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= testClassCreation "self debug: #testClassCreation" | mcls cls | mcls := MyMetaclass subclass: #MC instanceVariableNames: '' classVariableNames: '' poolDictionaries: '' category: 'MyMetaClassesGenerated'. cls := MyObject subclass: #C instanceVariableNames: '' classVariableNames: '' poolDictionaries: '' category: 'MyMetaClassesGenerated' metaclass: mcls. classCreated addAll: {mcls . cls}. "Used in teardown to remove the classes" mcls compile: 'foo ^20'. self assert: (cls new class == cls). self assert: (cls class name = 'C class'). self assert: (cls class class == mcls). self assert: (cls class class class class == MyMetaclass). self assert: (cls class soleInstance == cls). self assert: (cls canUnderstand: #foo) not. self assert: (cls class respondsTo: #foo). -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= Cheers, Alexandre -- _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: Alexandre Bergel http://www.cs.tcd.ie/Alexandre.Bergel ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. |
|
|
Hi,
Remember there is MetaclassTalk (http://csl.ensm-douai.fr/MetaclassTalk) that provides explicit metaclasses and other reflective facilities for Squeak. You can define your own metaclasses and use them in order to have classes with the following protocole : Object subclass: #Person instanceVariableNames: 'name age ' classVariableNames: '' poolDictionaries: '' category: 'MetaclassTalk-Examples' metaclass: MyMetaclass One intersting application of explicit metaclasses I did is to extend Smalltalk with mixin-based inheritance. However, neither implicit metaclasses (with parallel hierarchies) nor explicit metaclasses are totally satisfactory. Actually, there should be a mix of both (see my papers on metaclass compatibility OOPSLA98, ESUG2003). Noury Le 16 févr. 06, à 19:37, Bergel, Alexandre a écrit : > Today I spend some time to read and understand an old mail from > Andreas > (http://lists.squeakfoundation.org/pipermail/squeak-dev/2005-October/ > 096370.html) about having a parallel metaclass hierarchy. I really > like this. > > Thanks Andreas!! > > I wrote a short test to understand it, I think it might interest some > of you: > -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= > testClassCreation > "self debug: #testClassCreation" > | mcls cls | > > mcls := MyMetaclass > subclass: #MC > instanceVariableNames: '' > classVariableNames: '' > poolDictionaries: '' > category: 'MyMetaClassesGenerated'. > cls := MyObject > subclass: #C > instanceVariableNames: '' > classVariableNames: '' > poolDictionaries: '' > category: 'MyMetaClassesGenerated' > metaclass: mcls. > classCreated addAll: {mcls . cls}. "Used in teardown to remove the > classes" > > mcls compile: 'foo ^20'. > > self assert: (cls new class == cls). > self assert: (cls class name = 'C class'). > self assert: (cls class class == mcls). > self assert: (cls class class class class == MyMetaclass). > self assert: (cls class soleInstance == cls). > > self assert: (cls canUnderstand: #foo) not. > self assert: (cls class respondsTo: #foo). > -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= > > Cheers, > Alexandre > -- > _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: > Alexandre Bergel http://www.cs.tcd.ie/Alexandre.Bergel > ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. > > > > Dr. Noury Bouraqadi - Enseignant/Chercheur Ecole des Mines de Douai - Dept. G.I.P http://csl.ensm-douai.fr/noury European Smalltalk Users Group Board http://www.esug.org Squeak: an Open Source Smalltalk http://www.squeak.org -------------------------------------------------------------- |
|
|
According to the wiki page, Squeak 3.2 seems to be the platform used
for Metaclasstalk... Do you have the changeset somewhere ? Will it work with squeak 7000 ? Traits showed it should be difficult to make a .mcz file for this kind of think... Cheers, Alexandre Am Feb 20, 2006 um 9:16 PM schrieb Noury Bouraqadi: > Hi, > > Remember there is MetaclassTalk (http://csl.ensm-douai.fr/ > MetaclassTalk) that provides explicit metaclasses and other > reflective facilities for Squeak. You can define your own > metaclasses and use them in order to have classes with the > following protocole : > > Object subclass: #Person > instanceVariableNames: 'name age ' > classVariableNames: '' > poolDictionaries: '' > category: 'MetaclassTalk-Examples' > metaclass: MyMetaclass > > One intersting application of explicit metaclasses I did is to > extend Smalltalk with mixin-based inheritance. > However, neither implicit metaclasses (with parallel hierarchies) > nor explicit metaclasses are totally satisfactory. Actually, there > should be a mix of both (see my papers on metaclass compatibility > OOPSLA98, ESUG2003). > > Noury > Le 16 févr. 06, à 19:37, Bergel, Alexandre a écrit : > >> Today I spend some time to read and understand an old mail from >> Andreas (http://lists.squeakfoundation.org/pipermail/squeak-dev/ >> 2005-October/096370.html) about having a parallel metaclass >> hierarchy. I really like this. >> >> Thanks Andreas!! >> >> I wrote a short test to understand it, I think it might interest >> some of you: >> -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= >> testClassCreation >> "self debug: #testClassCreation" >> | mcls cls | >> >> mcls := MyMetaclass >> subclass: #MC >> instanceVariableNames: '' >> classVariableNames: '' >> poolDictionaries: '' >> category: 'MyMetaClassesGenerated'. >> cls := MyObject >> subclass: #C >> instanceVariableNames: '' >> classVariableNames: '' >> poolDictionaries: '' >> category: 'MyMetaClassesGenerated' >> metaclass: mcls. >> classCreated addAll: {mcls . cls}. "Used in teardown to remove >> the classes" >> >> mcls compile: 'foo ^20'. >> >> self assert: (cls new class == cls). >> self assert: (cls class name = 'C class'). >> self assert: (cls class class == mcls). >> self assert: (cls class class class class == MyMetaclass). >> self assert: (cls class soleInstance == cls). >> >> self assert: (cls canUnderstand: #foo) not. >> self assert: (cls class respondsTo: #foo). >> -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= >> >> Cheers, >> Alexandre >> -- >> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: >> Alexandre Bergel http://www.cs.tcd.ie/Alexandre.Bergel >> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. >> >> >> >> > -------------------------------------------------------------- > Dr. Noury Bouraqadi - Enseignant/Chercheur > Ecole des Mines de Douai - Dept. G.I.P > http://csl.ensm-douai.fr/noury > > European Smalltalk Users Group Board > http://www.esug.org > > Squeak: an Open Source Smalltalk > http://www.squeak.org > -------------------------------------------------------------- > > > -- _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: Alexandre Bergel http://www.cs.tcd.ie/Alexandre.Bergel ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. |
|
|
On 20 févr. 06, at 23:36, Alexandre Bergel wrote: > According to the wiki page, Squeak 3.2 seems to be the platform > used for Metaclasstalk... > Do you have the changeset somewhere ? Will it work with squeak 7000 ? > Traits showed it should be difficult to make a .mcz file for this > kind of think... Why? They should be orthogonal can you elaborate? |
|
|
> Why? They should be orthogonal
> can you elaborate? I guess Metaclasstalk needs a kind of bootstrap mechanism. Which are difficult to package with a .mcz. I played a bit with a new metaclass hierarchy, and I had problem to have a .mcz file. And what I did was simple... I am not saying that metaclasstalk interferes with traits. I am just saying that as it is difficult to produce a .mcz for traits, it is (probably) difficult to produce a .mcz for metaclasstalk. Alexandre -- _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: Alexandre Bergel http://www.cs.tcd.ie/Alexandre.Bergel ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. |
|
|
Actually, I didn't port MetaclassTalk because I'm lacking time to do
it. However, if you're intersted in it, I can help you. Regarding monticello and bootstrapping, I guess this can be done in some class initialize method. However, conceptually speaking, I don't like that class initialize method be used for package initializiation. It looks to me more like a hack to workarround a weakness of MC. Noury Le 21 févr. 06, à 16:43, Alexandre Bergel a écrit : >> Why? They should be orthogonal >> can you elaborate? > > I guess Metaclasstalk needs a kind of bootstrap mechanism. Which are > difficult to package with a .mcz. > I played a bit with a new metaclass hierarchy, and I had problem to > have a .mcz file. And what I did was simple... > > I am not saying that metaclasstalk interferes with traits. I am just > saying that as it is difficult to produce a .mcz for traits, it is > (probably) difficult to produce a .mcz for metaclasstalk. > > Alexandre > > -- > _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: > Alexandre Bergel http://www.cs.tcd.ie/Alexandre.Bergel > ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. > > > > > Dr. Noury Bouraqadi - Enseignant/Chercheur Ecole des Mines de Douai - Dept. G.I.P http://csl.ensm-douai.fr/noury European Smalltalk Users Group Board http://www.esug.org Squeak: an Open Source Smalltalk http://www.squeak.org -------------------------------------------------------------- |
«
Return to Squeak - Dev
|
1 view|%1 views
| Free forum by Nabble | Edit this page |