November 2013 - Pharo-users

Re: [Pharo-users] Lock free structures in Pharo

by jtuchel＠objektfabrik.de

Hi Norbert, I couldn't have said it any better. For me, this also was interesting, because I first thought: "Heck, how would I make anything atomic anyways?". And that question still stands: is there a way to make something atomic that includes message sends? I guess #critical is not the solution, is it? Joachim Am 14.11.13 11:12, schrieb Norbert Hartl: > Am 14.11.2013 um 10:25 schrieb Marcus Denker <marcus.denker(a)inria.fr>: > >> On 14 Nov 2013, at 10:15, Stephan Eggermont <stephan(a)stack.nl> wrote: >> >>> Reading this code, made me wonder what operations are actually atomic. >>> Anyone having a good explanation? >>> >>> Stephan >>> >>> AtomicQueueItem>makeCircular >>> "Make a receiver circular, i.e. point to itself, >>> answer the old value of next variable. >>> Note, this operation should be atomic" >>> >>> | temp | >>> >>> " atomic swap here" >>> temp := next. >>> next := self. >>> >>> ^ temp >>> >> -> no message send >> -> no back jump bytecode >> >> therefore it can not be interrupted and process switches can not happen between the statements. > Thanks, I learn something new every day. Iâve never thought about the condition when a process switch can happen. Can you say in short when a switch of processes is checked? Because I think I wonât finding it in a reasonable time looking myself. > > Norbert > > > -- ----------------------------------------------------------------------- Objektfabrik Joachim Tuchel mailto:jtuchel@objektfabrik.de Fliederweg 1 http://www.objektfabrik.de D-71640 Ludwigsburg http://joachimtuchel.wordpress.com Telefon: +49 7141 56 10 86 0 Fax: +49 7141 56 10 86 1

Nov. 14, 2013

Re: [Pharo-users] Lock free structures in Pharo

by Norbert Hartl

Am 14.11.2013 um 10:25 schrieb Marcus Denker <marcus.denker(a)inria.fr>: > > On 14 Nov 2013, at 10:15, Stephan Eggermont <stephan(a)stack.nl> wrote: > >> Reading this code, made me wonder what operations are actually atomic. >> Anyone having a good explanation? >> >> Stephan >> >> AtomicQueueItem>makeCircular >> "Make a receiver circular, i.e. point to itself, >> answer the old value of next variable. >> Note, this operation should be atomic" >> >> | temp | >> >> " atomic swap here" >> temp := next. >> next := self. >> >> ^ temp >> > -> no message send > -> no back jump bytecode > > therefore it can not be interrupted and process switches can not happen between the statements. Thanks, I learn something new every day. Iâve never thought about the condition when a process switch can happen. Can you say in short when a switch of processes is checked? Because I think I wonât finding it in a reasonable time looking myself. Norbert

Nov. 14, 2013

Re: [Pharo-users] Who uses poolDictionaries? New class template cleanup

by jtuchel＠objektfabrik.de

Yuri, Okay, so why not go one step further and kill PoolDictionaries? I mean, if no one uses them and you'd like to hide them from the users, they obviously are unnecessary. That would be a real cleanup, right? PoolDictionaries are potentially dangerous, because you can put things there that make Images harder to reproduce if you don't put the code to fill the Dictionaries into some script that will be run in the right situations (places like #loaded in envy). Joachim Am 14.11.13 11:05, schrieb Yuriy Tymchuk: > On 14 Nov 2013, at 11:02, jtuchel(a)objektfabrik.de wrote: > >> Hi, >> >> I'd like to draw your attention to something else: consistency. >> >> I remember the days when Cincom added Namespaces to classes and the class definition template changed drastically. It took me quite a while to get used to that. >> And there is this other aspect in the consistency field: most Smalltalk literature will include and address that line. If you learn Smalltalk, such small differences can cause trouble. > Is anything mentioning PoolDictionaries? I have no idea what it is. > > The other thing is that we should go from text for class creation to nice ui, but itâs a long term goal. > >> And what if I need PoolDictionaries? How hard will it be to find the place to add my reference to them? >> A fair amount of Pharo code may not make much use of PoolDictionaries, but other dialects do. I know Pharo has the concept of "never look back" and it is good to be prepared to cut off old strings. But there is a price to it. >> >> Seaside, apart from being a great web framework, has achieved something that was excellent and helped the whole Smalltalk universe make a leap forward: all Smalltalk dialects moved closer together and honestly worked on being more compatible. Suggestions like this may not break much of this per se, but many such cracks make a wide canyon. I find the argument that a certain line of code may irritate students a bit weak. It may make their life harder once they read code from other dialects. Should we remove class browsers because students are used to use text editors? >> >> Just my 2 cents >> >> Joachim >> >> Am 14.11.13 10:49, schrieb Martin Dias: >>> +10 >>> >>> >>> On Tue, Nov 12, 2013 at 9:24 PM, StÃ©phane Ducasse >>> <stephane.ducasse(a)inria.fr> wrote: >>>> +1 >>>> >>>> On Nov 12, 2013, at 4:39 PM, Johan Fabry <jfabry(a)dcc.uchile.cl> wrote: >>>> >>>>> Hi all, >>>>> >>>>> I am preparing slides for a course. I came to the typical: >>>>> >>>>> Number subclass: #Complex >>>>> instanceVariableNames: 'real imaginary' >>>>> classVariableNames: '' >>>>> poolDictionaries: '' >>>>> category: 'ComplexNumbers' >>>>> >>>>> Which made me think: Who uses poolDictionaries? I suspect extremely few of us. >>>>> >>>>> Why not add this to Class: >>>>> Class>>subclass: aSymbol instanceVariableNames: instVarNames classVariableNames: classVarNames category: aSymbol >>>>> ^self subclass: aSymbol >>>>> instanceVariableNames: instVarNames >>>>> classVariableNames: classVarNames >>>>> poolDictionaries: '' >>>>> category: aSymbol >>>>> >>>>> And have the new class template as follows? >>>>> >>>>> Object subclass: #NameOfSubclass >>>>> instanceVariableNames: '' >>>>> classVariableNames: '' >>>>> category: 'Kernel-Classes' >>>>> >>>>> It would be a bit cleaner. I know us old timers don't even see the poolDictionaries: line anymore, but I dislike having to explain it to students. >>>>> >>>>> Greetings, >>>>> >>>>> ---> Save our in-boxes! http://emailcharter.org <--- >>>>> >>>>> Johan Fabry - http://pleiad.cl/~jfabry >>>>> PLEIAD lab - Computer Science Department (DCC) - University of Chile >>>>> >>>>> >> >> -- >> ----------------------------------------------------------------------- >> Objektfabrik Joachim Tuchel mailto:jtuchel@objektfabrik.de >> Fliederweg 1 http://www.objektfabrik.de >> D-71640 Ludwigsburg http://joachimtuchel.wordpress.com >> Telefon: +49 7141 56 10 86 0 Fax: +49 7141 56 10 86 1 >> >> > > -- ----------------------------------------------------------------------- Objektfabrik Joachim Tuchel mailto:jtuchel@objektfabrik.de Fliederweg 1 http://www.objektfabrik.de D-71640 Ludwigsburg http://joachimtuchel.wordpress.com Telefon: +49 7141 56 10 86 0 Fax: +49 7141 56 10 86 1

Nov. 14, 2013

Re: [Pharo-users] Who uses poolDictionaries? New class template cleanup

by Yuriy Tymchuk

On 14 Nov 2013, at 11:02, jtuchel(a)objektfabrik.de wrote: > Hi, > > I'd like to draw your attention to something else: consistency. > > I remember the days when Cincom added Namespaces to classes and the class definition template changed drastically. It took me quite a while to get used to that. > And there is this other aspect in the consistency field: most Smalltalk literature will include and address that line. If you learn Smalltalk, such small differences can cause trouble. Is anything mentioning PoolDictionaries? I have no idea what it is. The other thing is that we should go from text for class creation to nice ui, but itâs a long term goal. > > And what if I need PoolDictionaries? How hard will it be to find the place to add my reference to them? > A fair amount of Pharo code may not make much use of PoolDictionaries, but other dialects do. I know Pharo has the concept of "never look back" and it is good to be prepared to cut off old strings. But there is a price to it. > > Seaside, apart from being a great web framework, has achieved something that was excellent and helped the whole Smalltalk universe make a leap forward: all Smalltalk dialects moved closer together and honestly worked on being more compatible. Suggestions like this may not break much of this per se, but many such cracks make a wide canyon. I find the argument that a certain line of code may irritate students a bit weak. It may make their life harder once they read code from other dialects. Should we remove class browsers because students are used to use text editors? > > Just my 2 cents > > Joachim > > Am 14.11.13 10:49, schrieb Martin Dias: >> +10 >> >> >> On Tue, Nov 12, 2013 at 9:24 PM, StÃ©phane Ducasse >> <stephane.ducasse(a)inria.fr> wrote: >>> +1 >>> >>> On Nov 12, 2013, at 4:39 PM, Johan Fabry <jfabry(a)dcc.uchile.cl> wrote: >>> >>>> Hi all, >>>> >>>> I am preparing slides for a course. I came to the typical: >>>> >>>> Number subclass: #Complex >>>> instanceVariableNames: 'real imaginary' >>>> classVariableNames: '' >>>> poolDictionaries: '' >>>> category: 'ComplexNumbers' >>>> >>>> Which made me think: Who uses poolDictionaries? I suspect extremely few of us. >>>> >>>> Why not add this to Class: >>>> Class>>subclass: aSymbol instanceVariableNames: instVarNames classVariableNames: classVarNames category: aSymbol >>>> ^self subclass: aSymbol >>>> instanceVariableNames: instVarNames >>>> classVariableNames: classVarNames >>>> poolDictionaries: '' >>>> category: aSymbol >>>> >>>> And have the new class template as follows? >>>> >>>> Object subclass: #NameOfSubclass >>>> instanceVariableNames: '' >>>> classVariableNames: '' >>>> category: 'Kernel-Classes' >>>> >>>> It would be a bit cleaner. I know us old timers don't even see the poolDictionaries: line anymore, but I dislike having to explain it to students. >>>> >>>> Greetings, >>>> >>>> ---> Save our in-boxes! http://emailcharter.org <--- >>>> >>>> Johan Fabry - http://pleiad.cl/~jfabry >>>> PLEIAD lab - Computer Science Department (DCC) - University of Chile >>>> >>>> >>> >> > > > -- > ----------------------------------------------------------------------- > Objektfabrik Joachim Tuchel mailto:jtuchel@objektfabrik.de > Fliederweg 1 http://www.objektfabrik.de > D-71640 Ludwigsburg http://joachimtuchel.wordpress.com > Telefon: +49 7141 56 10 86 0 Fax: +49 7141 56 10 86 1 > >

Nov. 14, 2013

Re: [Pharo-users] Who uses poolDictionaries? New class template cleanup

by jtuchel＠objektfabrik.de

Hi, I'd like to draw your attention to something else: consistency. I remember the days when Cincom added Namespaces to classes and the class definition template changed drastically. It took me quite a while to get used to that. And there is this other aspect in the consistency field: most Smalltalk literature will include and address that line. If you learn Smalltalk, such small differences can cause trouble. And what if I need PoolDictionaries? How hard will it be to find the place to add my reference to them? A fair amount of Pharo code may not make much use of PoolDictionaries, but other dialects do. I know Pharo has the concept of "never look back" and it is good to be prepared to cut off old strings. But there is a price to it. Seaside, apart from being a great web framework, has achieved something that was excellent and helped the whole Smalltalk universe make a leap forward: all Smalltalk dialects moved closer together and honestly worked on being more compatible. Suggestions like this may not break much of this per se, but many such cracks make a wide canyon. I find the argument that a certain line of code may irritate students a bit weak. It may make their life harder once they read code from other dialects. Should we remove class browsers because students are used to use text editors? Just my 2 cents Joachim Am 14.11.13 10:49, schrieb Martin Dias: > +10 > > > On Tue, Nov 12, 2013 at 9:24 PM, StÃ©phane Ducasse > <stephane.ducasse(a)inria.fr> wrote: >> +1 >> >> On Nov 12, 2013, at 4:39 PM, Johan Fabry <jfabry(a)dcc.uchile.cl> wrote: >> >>> Hi all, >>> >>> I am preparing slides for a course. I came to the typical: >>> >>> Number subclass: #Complex >>> instanceVariableNames: 'real imaginary' >>> classVariableNames: '' >>> poolDictionaries: '' >>> category: 'ComplexNumbers' >>> >>> Which made me think: Who uses poolDictionaries? I suspect extremely few of us. >>> >>> Why not add this to Class: >>> Class>>subclass: aSymbol instanceVariableNames: instVarNames classVariableNames: classVarNames category: aSymbol >>> ^self subclass: aSymbol >>> instanceVariableNames: instVarNames >>> classVariableNames: classVarNames >>> poolDictionaries: '' >>> category: aSymbol >>> >>> And have the new class template as follows? >>> >>> Object subclass: #NameOfSubclass >>> instanceVariableNames: '' >>> classVariableNames: '' >>> category: 'Kernel-Classes' >>> >>> It would be a bit cleaner. I know us old timers don't even see the poolDictionaries: line anymore, but I dislike having to explain it to students. >>> >>> Greetings, >>> >>> ---> Save our in-boxes! http://emailcharter.org <--- >>> >>> Johan Fabry - http://pleiad.cl/~jfabry >>> PLEIAD lab - Computer Science Department (DCC) - University of Chile >>> >>> >> > -- ----------------------------------------------------------------------- Objektfabrik Joachim Tuchel mailto:jtuchel@objektfabrik.de Fliederweg 1 http://www.objektfabrik.de D-71640 Ludwigsburg http://joachimtuchel.wordpress.com Telefon: +49 7141 56 10 86 0 Fax: +49 7141 56 10 86 1

Nov. 14, 2013

Re: [Pharo-users] Who uses poolDictionaries? New class template cleanup

by Martin Dias

+10 On Tue, Nov 12, 2013 at 9:24 PM, StÃ©phane Ducasse <stephane.ducasse(a)inria.fr> wrote: > +1 > > On Nov 12, 2013, at 4:39 PM, Johan Fabry <jfabry(a)dcc.uchile.cl> wrote: > >> Hi all, >> >> I am preparing slides for a course. I came to the typical: >> >> Number subclass: #Complex >> instanceVariableNames: 'real imaginary' >> classVariableNames: '' >> poolDictionaries: '' >> category: 'ComplexNumbers' >> >> Which made me think: Who uses poolDictionaries? I suspect extremely few of us. >> >> Why not add this to Class: >> Class>>subclass: aSymbol instanceVariableNames: instVarNames classVariableNames: classVarNames category: aSymbol >> ^self subclass: aSymbol >> instanceVariableNames: instVarNames >> classVariableNames: classVarNames >> poolDictionaries: '' >> category: aSymbol >> >> And have the new class template as follows? >> >> Object subclass: #NameOfSubclass >> instanceVariableNames: '' >> classVariableNames: '' >> category: 'Kernel-Classes' >> >> It would be a bit cleaner. I know us old timers don't even see the poolDictionaries: line anymore, but I dislike having to explain it to students. >> >> Greetings, >> >> ---> Save our in-boxes! http://emailcharter.org <--- >> >> Johan Fabry - http://pleiad.cl/~jfabry >> PLEIAD lab - Computer Science Department (DCC) - University of Chile >> >> > >

Nov. 14, 2013

Re: [Pharo-users] Lock free structures in Pharo

by Marcus Denker

On 14 Nov 2013, at 10:15, Stephan Eggermont <stephan(a)stack.nl> wrote: > Reading this code, made me wonder what operations are actually atomic. > Anyone having a good explanation? > > Stephan > > AtomicQueueItem>makeCircular > "Make a receiver circular, i.e. point to itself, > answer the old value of next variable. > Note, this operation should be atomic" > > | temp | > > " atomic swap here" > temp := next. > next := self. > > ^ temp > -> no message send -> no back jump bytecode therefore it can not be interrupted and process switches can not happen between the statements. Marcus

Nov. 14, 2013

Lock free structures in Pharo

by Stephan Eggermont

Reading this code, made me wonder what operations are actually atomic. Anyone having a good explanation? Stephan AtomicQueueItem>makeCircular "Make a receiver circular, i.e. point to itself, answer the old value of next variable. Note, this operation should be atomic" | temp | " atomic swap here" temp := next. next := self. ^ temp

Nov. 14, 2013

Re: [Pharo-users] NBOpenGL on Pharo 3.0

by Richard Wettel

Thanks everyone! That would be great. Looking forward to having NBOpenGL up and running again. Cheers Ricky On Thu, Nov 14, 2013 at 2:29 AM, StÃ©phane Ducasse <stephane.ducasse(a)inria.fr > wrote: > >> Right now it cannot even be loaded into 3.0. > But update is simple (i think JB did that already, just not published?). > > Grrr, all my code are published ^^. > > > Where? When did you announce it? > Sorry but how can you expect guys from the other part of the planet to > know it. > Communication in open-source is important because people do not know what > others > are doing. > You know richie asked and ronie too and they are smart guys. But they > cannot guess. > > For mac: > I do the change for mac last month because I needed it, it is integrated. > > > ah ok this is why it is working on mac. > > For linux: > I fix, published, tested and updated the configuration for linux 2 days > ago, because Alex ask me. > On my roommate computer that work (ubuntu ). > But I need feedback. > > > oki we will check that tomorrow I hope. > > For windows: > I publish a fix for window yesterday:NBOpenGL-Win-jeanbaptistearnaud.9 but > I do not update configuration because I cannot test it. > I was planned to go test and debug on my home computer (window 7, 64), > tomorrow. > > > Ok we can say it to ronie to have a look. > > So, you can expect a fix integrated soon. > > And having a working OpenGL, exactly as it work in pharo 2.0. > > > Excellent. > > > > > >> I have already looked at Roassal 3d, but it was not easy to separate the >> API from the code. >> >> >> Strange because even me I could see that the shadder code should not be >> in Roassal 3D but in NBOpenGL :). >> Because Roassal 3d should not be about shadder but about 3D. >> > >> And it is something completely different than what I have been using (and >> SourceCity, too), because Roassal 3d is using the modern OpenGL methodology >> (with FBOs and stuff). >> >> >> Yes they changed the API and deprecated the old way of doing it. >> > > I will also publish some example (using shadder) that I wrote but it is > still not yet user friendly. > > > So, I would really appreciate if that API would find its way from Roassal >> to NBOpenGL and I could use it for CodeCity. >> >> >> See below >> > >> Unfortunately I am able to help a lot with that, because I know very >> little about this and I also don't have enough time (I am doing CodeCity in >> my spare time). But, if you have concrete things that I can do, I'd be >> happy to help. >> >> >> Three remarks: >> - I know that jean-baptiste worked on moving part of roassal3d to NBOpenGL >> > > Shader part are done for now. > > - I'm not sure that doing 3d without investing a bit in the underlying >> technology is wise. Because you will be always relying >> on other people and probably frustrated when things are not working. >> > + 1 > > - Since JB does not know how to communicate here is what he did for fun >> instead of pushing MoosePython :) >> > a renderer to video to videos stream for openGL -> MKV, mp4 based one >> ffmpeg). The code is openGL extra. As an example you >> can get >> https://www.dropbox.com/sh/uyli3tx3drendvj/qzA_Hy0jnk \ >> > >> > yes, my bad. > > Now again if nobody builds a jun like library in Pharo it will not exist. >> > >> Stef >> >> >> >> Cheers >> Ricky >> >> >> >> >> On Wed, Nov 13, 2013 at 9:44 AM, Tudor Girba <tudor(a)tudorgirba.com>wrote: >> >>> Hi, >>> >>> I guess that what Ricky is asking is what changed between NBOpenGL 2.0 >>> and 3.0 such that this does not work anymore on Windows. The fact that he >>> mentioned 64 bits is not relevant for this discussion :). >>> >>> I also guess he wants to put some effort into understanding this as well >>> but he needs a bit of guidance. Right Ricky? >>> >>> Cheers, >>> Doru >>> >>> >>> >>> On Tue, Nov 12, 2013 at 2:12 PM, StÃ©phane Ducasse < >>> stephane.ducasse(a)inria.fr> wrote: >>> >>>> Richard >>>> >>>> you should have a look at ROASSAL 3d because ronie improved the shadder >>>> part (no more assembler) >>>> and his code should be split in two and one part should go to NBOpenGL. >>>> Now he is experiencing some problem with NBOpenGL on windows as you. >>>> >>>> NativeBoost does not work on 64 bits. Yet but igor is really busy >>>> finishing the texteditor (and I can tell you that >>>> he does not have fun there). >>>> >>>> So it is important that you help because we are FULL >>>> and sorry about that but really FULL. >>>> >>>> Ronie will visit us in January and share an office with igor. >>>> >>>> Stef >>>> >>>> >>>> > Hi >>>> > >>>> > CodeCity, the project I am working on is based on NBOpenGL. >>>> > I'd like to start working on it using Pharo 3, but it seems that >>>> NBOpenGL does not work yet with Pharo 3.0. >>>> > >>>> > I downloaded from jenkins the latest image with Pharo 3 and tried to >>>> execute the following code on Windows 7, 64 bits: >>>> > >>>> > GLTTRenderingDemo new openInWorld. >>>> > >>>> > but I get a 'No suitable implementation found for initializing OpenGL >>>> context for your platform'. >>>> > It seems that NBGLContextDriver does not have a subclass for Windows >>>> 64bits. Did I omit something? >>>> > >>>> > Cheers >>>> > Ricky >>>> > >>>> >>>> >>>> >>> >>> >>> -- >>> www.tudorgirba.com >>> >>> "Every thing has its own flow" >>> >> >> >> > > > -- > Best regards, > Igor Stasenko. > > > Best Regards > Jean Baptiste Arnaud > jbaptiste.arnaud(a)gmail.com > > > > > > > > >

Nov. 14, 2013

Re: [Pharo-users] Thread safe queue like collection

by GÃ¶ran Krampe

On 11/13/2013 08:30 PM, Sven Van Caekenberghe wrote: > Yes, I have used it, even extended it. > It looks thread safe, so it should be enough to get started. If you are after something that is more like an in memory SocketStream then you have the SharedStreams package which I wrote a long time ago, it is actually pretty cool: http://map.squeak.org/packagebyname/sharedstreams I recall that we did find several "holes" in SharedQueue at the time. SharedStreams also use more fitting prims for characters/bytes. regards, GÃ¶ran

Nov. 14, 2013