April 2012 - Pharo-dev - lists.pharo.org

[Pharo-project] gamedev.net post asking about Smalltalk

by askoh

Some is interested in using Smalltalk for prototyping of games. http://www.gamedev.net/topic/623380-game-programming-in-smalltalk/ I think we should respond very constructively on gamedev.net to entice gamers to try Smalltalk. Programmers are coming to appreciate dynamic languages and we can capitalize on that. Go Smalltalk, Aik-Siong Koh -- View this message in context: http://forum.world.st/gamedev-net-post-asking-about-Smalltalk-tp4559039p455… Sent from the Pharo Smalltalk mailing list archive at Nabble.com.

April 15, 2012

[Pharo-project] InternetConfigPlugin for unix - [WAS]Tips to create a plugin with specific platform dependant stuff

by Guillermo Polito

Bump :) I've been working last week a bit on the internet plugin to fetch some system proxy configuration in unix. Since - the plugin is external - it loads gconf libraries using dlopen, so the libraries are not needed during compilation nor running - if gconf library is not found, looks at the http-proxy and ftp-proxy env vars. It should break nothing :P But I'm not an expert :/. What do you think? Is it worth of integration? Guille On Mon, Apr 9, 2012 at 4:38 PM, Guillermo Polito <guillermopolito(a)gmail.com>wrote: > ok, > > I did this: > - configured the plugin as external > - created > > CogFamilyUnixConfig>>configureInternetConfigPlugin: maker > super configureInternetConfigPlugin: maker. > maker addPlatformSources: #( 'sqUnixInternetConfiguration') > > and put the file attached in /platforms/unix/plugins/InternetConfigPlugin. > I finally used dlopen to load gconf dinamically and check if it is > available, and if it's not, I check the env vars with getenv(). > > If someone can do a little review, it would be nice :). > > Guille > > > On Sun, Apr 8, 2012 at 9:55 PM, Guillermo Polito < > guillermopolito(a)gmail.com> wrote: > >> Haha! :) >> >> You won me :). I was about to reply this: >> >> "Ok, so I talked with Esteban today :). And thanks to him I reached the >> conclussion that this plugin should be external." >> >> I'll spend some hours tomorrow to finish it. >> >> Now, as I think it happens with the ssl plugin... Where should we put the >> code and what should I in order to get it easily integrated in the ci >> buildings? :) >> >> Thanks! >> Guille >> >> >> On Sun, Apr 8, 2012 at 9:46 PM, Igor Stasenko <siguctua(a)gmail.com> wrote: >> >>> On 7 April 2012 17:55, Guillermo Polito <guillermopolito(a)gmail.com> >>> wrote: >>> > Was playing in my weekend on the InternetConfigPlugin for ubuntu. And >>> > that's the problem :), It uses gnome2 stuff to access the system >>> > configuration (because gnome handles it's own configuration... :/). >>> > >>> > Now, I made it work doing something like this in vmmaker: >>> > >>> > CogFamilyUnixConfig>>configureInternetConfigPlugin: maker >>> > "extra rules for InternetConfigPlugin" >>> > >>> > super configureInternetConfigPlugin: maker. >>> > maker addDefinitions: '`pkg-config --libs --cflags gtk+-2.0 >>> gconf-2.0`'. >>> > maker addExternalLibrary: '/usr/lib/libgconf-2.so.4'. >>> > maker addPlatformSources: #( 'sqUnixInternetConfiguration.c'). >>> > >>> > And adding it in the list of internal plugins for Unix. >>> > >>> > But this should only compile and work on a system with gconf and stuff >>> > installed :). >>> > >>> Then don't make this plugin internal, but external instead. Because if >>> library is missing, >>> VM will refuse to start. In contrast, if your plugin will be in >>> external module, then module will refuse to load >>> if lib is missing. >>> >>> > So, how does or should vmmaker and vm building process handle >>> something like >>> > this? >>> > >>> >>> The philosophy of CMake configurations is to be concrete and without >>> conditionals. I.E. they should >>> not contain "if this, do this, if that do that" , instead if you have >>> such choice, one should make another configuration. >>> >>> There is no way to predict how different is platform on which you >>> building VM from platform where it used (especially in case of unix >>> systems and their numerous flavors). So, what is working on your OS, >>> could not work on another. That's why i don't see a point to put >>> conditionals in build process. >>> >>> In contrast, if you have a concrete configuration, which says: if you >>> want to build me, you should have this, this and this, and if you want >>> to use the built artifact you should have this, this and this >>> installed, then it makes things much more predictable and >>> straightforward. >>> >>> >>> > Guille >>> > >>> > _______________________________________________ >>> > VM-beginners mailing list >>> > VM-beginners(a)lists.squeakfoundation.org >>> > http://lists.squeakfoundation.org/mailman/listinfo/vm-beginners >>> > >>> >>> -- >>> Best regards, >>> Igor Stasenko. >>> >> >> >

April 15, 2012

Re: [Pharo-project] Pharo one click 1.4 and Linux COGVM

by Guillermo Polito

I've uploaded the fix to https://gitorious.org/cogvm/blessed Now, have to wait for the new build in here: https://ci.lille.inria.fr/pharo/job/Cog%20Git%20Tracker/ https://ci.lille.inria.fr/pharo/view/Cog/job/Cog-VM/Architecture=32,OS=linu… Guille On Sat, Apr 14, 2012 at 8:04 PM, Guillermo Polito <guillermopolito(a)gmail.com > wrote: > Cog vm mappings are broken for the numpad. I the fix, but I have to > integrate it. Maybe tomorrow :$. > > > On Sat, Apr 14, 2012 at 8:00 PM, Nahuel Garbezza <n.garbezza(a)gmail.com>wrote: > >> Hi, >> >> I'm having this is issue in 1.3 too... I always forget to ask why >> >> Nahuel. >> >> >> On 14 April 2012 10:27, Hilaire Fernandes <hilaire.fernandes(a)edu.ge.ch>wrote: >> >>> How stupid as it seems, numeric keyboard does not work under Linux and >>> several COG VM I tried on linux. Looks ok under Wine + Windows COG VM. >>> >>> How come? >>> >>> >>> -- >>> Dr. Geo -- http://www.drgeo.eu >>> >>> >>> >> >

April 15, 2012

Re: [Pharo-project] Stupid question about Fuel

by Marcus Denker

On Apr 15, 2012, at 2:11 PM, Edgar J. De Cleene wrote: > > > > On 4/15/12 8:50 AM, "Marcus Denker" <marcus.denker(a)inria.fr> wrote: > >> I would not waste time with that... we will improve Pharo constantly, and it >> will >> be very hard (if not impossible) to keep some kind of compatibility layer on a >> binary >> code level between Pharo, Squeak and Cuis. >> >> And what do you get out of it? Wouldn't it be better to use the time to >> improve one system >> and make sure that system has a future? >> >> Marcus > > And if none of actual forks have all as I wish? > All have some good and all have some problem for me. > It's why try to learn from all here and all in Squeak and from Juan. > Learning does not mean to be binary compatible. -- Marcus Denker -- http://marcusdenker.de

April 15, 2012

Re: [Pharo-project] Stupid question about Fuel

by Edgar J. De Cleene

On 4/15/12 8:50 AM, "Marcus Denker" <marcus.denker(a)inria.fr> wrote: > I would not waste time with that... we will improve Pharo constantly, and it > will > be very hard (if not impossible) to keep some kind of compatibility layer on a > binary > code level between Pharo, Squeak and Cuis. > > And what do you get out of it? Wouldn't it be better to use the time to > improve one system > and make sure that system has a future? > > Marcus And if none of actual forks have all as I wish? All have some good and all have some problem for me. It's why try to learn from all here and all in Squeak and from Juan. Don't' sell me the fish, teach me to fishing. Edgar

April 15, 2012

Re: [Pharo-project] Stupid question about Fuel

by Marcus Denker

On Apr 15, 2012, at 1:44 PM, Edgar J. De Cleene wrote: > > And wish made experiments if this Foo.fuel could be loaded into different > fork as Squeak, Cuis having Fuel into. > I would not waste time with that... we will improve Pharo constantly, and it will be very hard (if not impossible) to keep some kind of compatibility layer on a binary code level between Pharo, Squeak and Cuis. And what do you get out of it? Wouldn't it be better to use the time to improve one system and make sure that system has a future? Marcus -- Marcus Denker -- http://marcusdenker.de

April 15, 2012

Re: [Pharo-project] Stupid question about Fuel

by Edgar J. De Cleene

On 4/15/12 8:24 AM, "Marcus Denker" <marcus.denker(a)inria.fr> wrote: > Of course that can not work. .cs is source code, you of course need a compiler > (if Fuel could load cs.gz, it *would be* a compiler). Yes, I know it. I have the .cs.gz for normal load, not for Fuel !!! > What you need to do is to serialize the *Objects* that are code (Classes, > MethodDictioaries, > CompiledMethods) using Fuel. Then the result can be loaded back using Fuel. What I ask is how to have the .fuel of classes as regular objects as Mariano write in his blog. Then , this Foo.fuel could be loaded into target image. Some more granular as Pavel made with complete packages some days ago. And wish made experiments if this Foo.fuel could be loaded into different fork as Squeak, Cuis having Fuel into. > Marcus Thanks ! Edgar

April 15, 2012

[Pharo-project] Stupid question about Fuel

by Edgar J. De Cleene

Mariano, Martin, Pavel > For example, a class can be considered either a global or a regular object. In > the former case, it will be encoded just its name; in the latter case, the > class will be encoded in detail, with its method dictionary, etc. How you do this ? >the class will be encoded in detail, with its method dictionary, etc. I export most of 1.4 OneClick classes in http://squeakros.org/Pharo%201dot4/ as cs.gz, like to have .fuel for load classes back in target image with no compiler. Thanks in advance Edgar

April 15, 2012

Re: [Pharo-project] Stupid question about Fuel

by Marcus Denker

On Apr 15, 2012, at 1:20 PM, Edgar J. De Cleene wrote: > Mariano, Martin, Pavel > >> For example, a class can be considered either a global or a regular object. In >> the former case, it will be encoded just its name; in the latter case, the >> class will be encoded in detail, with its method dictionary, etc. > > How you do this ? >> the class will be encoded in detail, with its method dictionary, etc. > > I export most of 1.4 OneClick classes in http://squeakros.org/Pharo%201dot4/ > as cs.gz, like to have .fuel for load classes back in target image with no > compiler. Of course that can not work. .cs is source code, you of course need a compiler (if Fuel could load cs.gz, it *would be* a compiler). What you need to do is to serialize the *Objects* that are code (Classes, MethodDictioaries, CompiledMethods) using Fuel. Then the result can be loaded back using Fuel. Marcus -- Marcus Denker -- http://marcusdenker.de

April 15, 2012

Re: [Pharo-project] Some unicode initialization lost

by Guillermo Polito

http://code.google.com/p/pharo/issues/detail?can=2&q=5602&colspec=ID%20Type… On Sun, Apr 15, 2012 at 12:56 PM, Guillermo Polito < guillermopolito(a)gmail.com> wrote: > With this version of the method, now it works :). I've changed it to fill > the tables with the relative numbers it uses, just as the current > initialization. > > I'll open an issue. > > > On Sun, Apr 15, 2012 at 12:37 PM, Guillermo Polito < > guillermopolito(a)gmail.com> wrote: > >> Unicode class>>#parseUnicodeDataFrom: >> >> is the method in charge of initialize the unicode tables telling when a >> char is lowercase, or digit, etc.. >> >> But it has no senders :). >> >> Maybe a first try can be something like the case mappings: >> >> Unicode class>>initializeUnicodeData >> UIManager default informUserDuring: [ :bar| | result | >> bar value: 'Downloading Unicode data'. >> (result := ZnEasy get: ' >> http://www.unicode.org/Public/UNIDATA/UnicodeData.txt') isSuccess >> ifFalse: [ ^ self error: 'Download failed' ]. >> bar value: 'Updating Case Mappings'. >> self parseUnicodeDataFrom: result contents readStream ]. >> >> Now, is http://www.unicode.org/Public/UNIDATA/UnicodeData.txt the >> correct file to load the data from? >> >> BTW, I executed that code in my image and It broke into pieces, so the >> existing initialization may have some problem. Dunno what happens yet. >> >> Guille >> > >

April 15, 2012