December 2011 - Pharo-dev - lists.pharo.org

Re: [Pharo-project] Squeak/Pharo in Fedora

by Pavel Krivanek

On Tue, Dec 6, 2011 at 11:06 PM, Esteban Lorenzano <estebanlm(a)gmail.com> wrote: > Hi, > >> It is in hands of Esteban. Strange, I thought that the Cog development >> is not splitted. The Git version uses cmake. SVN version autotools? > > It is not... Igor first (and now myself) do an integration of what is on SVN regularly... also with a couple of patches of Igor not yet integrated to svn branch (AFAIK). > > We use gitorious because we believe is much better to community work the git way (svn is too restrictive), but we didn't split cog vm development. Right now, because the use of git and CMakeMaker, anyone can fork and do his own vm changes... then if they want they can ask for a merge... or not... I think is good for everybody. > > Said so... I do believe our versioning strategy is not good enough... is a little bit better now, but we need to improve. OK, thank you for the explanation. -- Pavel > > Cheers, > Esteban > >

Dec. 6, 2011

Re: [Pharo-project] Squeak/Pharo in Fedora

by Esteban Lorenzano

Hi, > It is in hands of Esteban. Strange, I thought that the Cog development > is not splitted. The Git version uses cmake. SVN version autotools? It is not... Igor first (and now myself) do an integration of what is on SVN regularly... also with a couple of patches of Igor not yet integrated to svn branch (AFAIK). We use gitorious because we believe is much better to community work the git way (svn is too restrictive), but we didn't split cog vm development. Right now, because the use of git and CMakeMaker, anyone can fork and do his own vm changes... then if they want they can ask for a merge... or not... I think is good for everybody. Said so... I do believe our versioning strategy is not good enough... is a little bit better now, but we need to improve. Cheers, Esteban

Dec. 6, 2011

Re: [Pharo-project] Squeak/Pharo in Fedora

by Eliot Miranda

On Tue, Dec 6, 2011 at 1:27 PM, Pavel Krivanek <pavel.krivanek(a)gmail.com>wrote: > On Tue, Dec 6, 2011 at 9:58 PM, Eliot Miranda <eliot.miranda(a)gmail.com> > wrote: > > > > > > On Tue, Dec 6, 2011 at 12:56 PM, Pavel Krivanek < > pavel.krivanek(a)gmail.com> > > wrote: > >> > >> Hi Eliot, > >> > >> On Tue, Dec 6, 2011 at 9:38 PM, Eliot Miranda <eliot.miranda(a)gmail.com> > >> wrote: > >> > > >> > > >> > On Tue, Dec 6, 2011 at 11:05 AM, Pavel Krivanek > >> > <pavel.krivanek(a)gmail.com> > >> > wrote: > >> >> > >> >> Hi, > >> >> > >> >> I want make a little report about progress with making of Fedora > >> >> pakcages for Smalltalk related projects (StackVM, CogVM, Pharo...) by > >> >> Jaroslav Å karvada. > >> >> > >> >> Currently there are three packages in Fedora repositories - > squeak-vm, > >> >> squeak-image and squeak-vm-nonXOplugins (that contains plugins like > >> >> OSProcess). It uses resources from the Squeak site. > >> >> > >> >> Red Hat has quite restrictive demands on packages and it is a small > >> >> miracle that they tolerate current squeak packages :-). The > conditions > >> >> simply do expect that a program is a set of source codes and not a > >> >> system of communicating objects. One problem is that Squeak VM is one > >> >> of few metacircular packages that are needed for their own building. > >> >> Because binary executables are prohibited, it would require a special > >> >> bureaucratically problematic exception. Fortunately for virtual > >> >> machine generation we may use the current squeak packages - more > >> >> precisely their updated versions. The next issue for virtual machine > >> >> generation may be the fact, that some packages used for VM generation > >> >> are in repositories with public read&write permissions. > >> >> > >> >> Because image can be likened to *.jar files that are prohibited and > >> >> must be recompiled during package generation, we are lucky that > >> >> squeak-image package is currently accepted. If someone starts > >> >> tinkering, we are in big trouble. > >> >> > >> >> There is a little issue with *.sources placement. The current VM > >> >> package contains symlinks from VM path to sources files that are > >> >> present in package with the image. This is really ugly solution. > >> >> Squeak and Pharo uses this sequence for sources lookup: vm path, > image > >> >> path, current path. So I think that the best solution will be to have > >> >> sources file in the image package, do not have symlinks in vm package > >> >> and to create symlink to sources together with changes in the image > >> >> path via mysqueak initialization script. > >> >> > >> >> Btw., Jaroslav told me that he saw a lot of creepy projects but > >> >> nothing like Squeak VM :-) > >> >> > >> >> And there is a mess in VM versioning. What is the latest > CogVM/StackVM > >> >> version number? > >> > > >> > > >> > How so? My latest version is SVN > >> > http://www.squeakvm.org/svn/squeak/branches/Cog/ r2519. Monticello > >> > package > >> > http://source.squeak.org/VMMaker/VMMaker.oscog-eem.139. > >> > >> What is the relation between this SVN repository and Git repository > >> used by the CI server (https://git.gitorious.org/cogvm/blessed.git)? > > > > > > Not anything to do with me. > > It is in hands of Esteban. Strange, I thought that the Cog development > is not splitted. The Git version uses cmake. SVN version autotools? > yes. > > >> What is the reason to exclude fbdev? > > > > > > What's fbdev? > > framebuffer display device. > It wasn't used at Qwaq/Teleplace and as things evolved it was easier to leave it behind. Do you want it back? > > >> > >> > >> -- Pavel > >> > >> >> Cheers, > >> >> -- Pavel > >> >> > >> > > >> > > >> > > >> > -- > >> > best, > >> > Eliot > >> > > >> > > > > > > > > -- > > best, > > Eliot > > > > -- best, Eliot

Dec. 6, 2011

Re: [Pharo-project] issue 5048 broke Pharo Kernel

by Pavel Krivanek

Great. Please add the new package name to kernelPackageNames too. -- Pavel On Tue, Dec 6, 2011 at 10:40 PM, StÃ©phane Ducasse <stephane.ducasse(a)inria.fr> wrote: > Ok I will do that. > Note that we started to remove systematically hard coded references to transcript so that we can even remove it. > We have std output now. > > Stef > > On Dec 6, 2011, at 6:50 PM, Pavel Krivanek wrote: > >> We need it in the Kernel because some classes log system messages >> there (like Compiler during recompilation of the image). I think that >> own separate package would be fine. >> >> -- Pavel >> >> On Tue, Dec 6, 2011 at 5:36 PM, StÃ©phane Ducasse >> <stephane.ducasse(a)inria.fr> wrote: >>> how can we fix it? >>> Because we believe that Transcript should not be packaged in Stream. >>> Now may be in its own package. >>> >>> Stef >>> >>> On Dec 6, 2011, at 3:40 PM, Pavel Krivanek wrote: >>> >>>> Hi, >>>> >>>> the issue http://code.google.com/p/pharo/issues/detail?id=5048 broke >>>> Pharo-Kernel build on CI server. >>>> >>>> -- Pavel >>>> >>> >>> >> > >

Dec. 6, 2011

Re: [Pharo-project] issue 5048 broke Pharo Kernel

by StÃ©phane Ducasse

Ok I will do that. Note that we started to remove systematically hard coded references to transcript so that we can even remove it. We have std output now. Stef On Dec 6, 2011, at 6:50 PM, Pavel Krivanek wrote: > We need it in the Kernel because some classes log system messages > there (like Compiler during recompilation of the image). I think that > own separate package would be fine. > > -- Pavel > > On Tue, Dec 6, 2011 at 5:36 PM, StÃ©phane Ducasse > <stephane.ducasse(a)inria.fr> wrote: >> how can we fix it? >> Because we believe that Transcript should not be packaged in Stream. >> Now may be in its own package. >> >> Stef >> >> On Dec 6, 2011, at 3:40 PM, Pavel Krivanek wrote: >> >>> Hi, >>> >>> the issue http://code.google.com/p/pharo/issues/detail?id=5048 broke >>> Pharo-Kernel build on CI server. >>> >>> -- Pavel >>> >> >> >

Dec. 6, 2011

Re: [Pharo-project] Squeak/Pharo in Fedora

by StÃ©phane Ducasse

can you open a ticket mariano? Tx On Dec 6, 2011, at 10:26 PM, Mariano Martinez Peck wrote: > > > How so? My latest version is SVN http://www.squeakvm.org/svn/squeak/branches/Cog/ r2519. Monticello package http://source.squeak.org/VMMaker/VMMaker.oscog-eem.139. > > But where is such information stored? > Then, the VMMaker package version says nothing, because to buiyld the VM you need more packages than VMMaker, so you have to guess which version to use. > Moreoever, there is also ANOTHER version in VM which is VMMaker class>>versionString which is different from that. > Then from the image or from a log we get: VM: Mac OS - intel - 1068 - Croquet Closure Cog VM [CoInterpreter VMMaker.oscog-eem.78] Croquet Cog 3.0.0 > so where it is the svn version? > > (1007 to: 1009) collect: [:n| Smalltalk getSystemAttribute: n] > > #('CoInterpreter VMMaker.oscog-eem.138 uuid: f9a63bdf-7bbc-4ae5-9634-ecd6fd814c9d Nov 18 2011' > 'StackToRegisterMappingCogit VMMaker.oscog-eem.139 uuid: c2849383-9768-4948-b9b2-a5c22d482b07 Nov 18 2011' > 'r2519 http://www.squeakvm.org/svn/squeak/branches/Cog') > > Good enough? > > Better. So I guess that at least in Pharo this method deserves an update: > > version > "Return the version number string of the Virtual machine built" > " > Smalltalk vm version > " > > ^(self getSystemAttribute: 1004), ' ', ((self getSystemAttribute: 1009) ifNil: ['']) > > > > > > In fact, only the last is required since the VM and plugin source is included in http://www.squeakvm.org/svn/squeak/branches/Cog/src. > > > So...all in all, I think that indeed, it is quite a mess. But it is also complicated to manage and I don't really know what to suggest. > With ConfigurationOfCog at least with one number we group all image side related packages. > Then we only need svn/git version. Moreover, we can store this svn/git as the description in each version of ConfigurationOfCog. That way, you know *exactly* what do you need. > > Cheers > > > > > Cheers, > -- Pavel > > > > > -- > best, > Eliot > > > > > -- > Mariano > http://marianopeck.wordpress.com > > > > > -- > best, > Eliot > > > > > -- > Mariano > http://marianopeck.wordpress.com >

Dec. 6, 2011

Re: [Pharo-project] Squeak/Pharo in Fedora

by Pavel Krivanek

On Tue, Dec 6, 2011 at 10:17 PM, Eliot Miranda <eliot.miranda(a)gmail.com> wrote: > > > On Tue, Dec 6, 2011 at 1:07 PM, Mariano Martinez Peck > <marianopeck(a)gmail.com> wrote: >> >> >> >> On Tue, Dec 6, 2011 at 9:38 PM, Eliot Miranda <eliot.miranda(a)gmail.com> >> wrote: >>> >>> >>> >>> On Tue, Dec 6, 2011 at 11:05 AM, Pavel Krivanek >>> <pavel.krivanek(a)gmail.com> wrote: >>>> >>>> Hi, >>>> >>>> I want make a little report about progress with making of Fedora >>>> pakcages for Smalltalk related projects (StackVM, CogVM, Pharo...) by >>>> Jaroslav Å karvada. >>>> >>>> Currently there are three packages in Fedora repositories - squeak-vm, >>>> squeak-image and squeak-vm-nonXOplugins (that contains plugins like >>>> OSProcess). It uses resources from the Squeak site. >>>> >>>> Red Hat has quite restrictive demands on packages and it is a small >>>> miracle that they tolerate current squeak packages :-). The conditions >>>> simply do expect that a program is a set of source codes and not a >>>> system of communicating objects. One problem is that Squeak VM is one >>>> of few metacircular packages that are needed for their own building. >>>> Because binary executables are prohibited, it would require a special >>>> bureaucratically problematic exception. Fortunately for virtual >>>> machine generation we may use the current squeak packages - more >>>> precisely their updated versions. The next issue for virtual machine >>>> generation may be the fact, that some packages used for VM generation >>>> are in repositories with public read&write permissions. >>>> >>>> Because image can be likened to *.jar files that are prohibited and >>>> must be recompiled during package generation, we are lucky that >>>> squeak-image package is currently accepted. If someone starts >>>> tinkering, we are in big trouble. >>>> >>>> There is a little issue with *.sources placement. The current VM >>>> package contains symlinks from VM path to sources files that are >>>> present in package with the image. This is really ugly solution. >>>> Squeak and Pharo uses this sequence for sources lookup: vm path, image >>>> path, current path. So I think that the best solution will be to have >>>> sources file in the image package, do not have symlinks in vm package >>>> and to create symlink to sources together with changes in the image >>>> path via mysqueak initialization script. >>>> >>>> Btw., Jaroslav told me that he saw a lot of creepy projects but >>>> nothing like Squeak VM :-) >>>> >>>> And there is a mess in VM versioning. What is the latest CogVM/StackVM >>>> version number? >>> >>> >>> How so? Â My latest version is SVN >>> http://www.squeakvm.org/svn/squeak/branches/Cog/ r2519. Â Monticello package >>> http://source.squeak.org/VMMaker/VMMaker.oscog-eem.139. >> >> >> But where is such information stored? >> Then, the VMMaker package version says nothing, because to buiyld the VM >> you need more packages than VMMaker, so you have to guess which version to >> use. >> Moreoever, there is also ANOTHER version in VM which is VMMaker >> class>>versionStringÂ which is different from that. >> Then from the image or from a log we get: VM: Mac OS - intel - 1068 - >> Croquet Closure Cog VM [CoInterpreter VMMaker.oscog-eem.78] Croquet Cog >> 3.0.0 >> so where it is the svn version? > > > (1007 to: 1009) collect: [:n| Smalltalk getSystemAttribute: n] > > #('CoInterpreter VMMaker.oscog-eem.138 uuid: > f9a63bdf-7bbc-4ae5-9634-ecd6fd814c9d Nov 18 2011' > Â Â 'StackToRegisterMappingCogit VMMaker.oscog-eem.139 uuid: > c2849383-9768-4948-b9b2-a5c22d482b07 Nov 18 2011' > Â Â 'r2519 http://www.squeakvm.org/svn/squeak/branches/Cog') > > Good enough? Â In fact, only the last is required since the VM and plugin > source is included inÂ http://www.squeakvm.org/svn/squeak/branches/Cog/src. > The VM originally used major.minor.revision naming. He was confused that he cannot see sequel. -- Pavel >> >> So...all in all, I think that indeed, it is quite a mess. But it is also >> complicated to manage and I don't really know what to suggest. >> With ConfigurationOfCog at least with one number we group all image side >> related packages. >> Then we only need svn/git version. Moreover, we can store this svn/git as >> the description in each version of ConfigurationOfCog. That way, you know >> *exactly* what do you need. >> >> Cheers >> >> >>> >>> >>>> >>>> >>>> Cheers, >>>> -- Pavel >>>> >>> >>> >>> >>> -- >>> best, >>> Eliot >>> >> >> >> >> -- >> Mariano >> http://marianopeck.wordpress.com >> > > > > -- > best, > Eliot >

Dec. 6, 2011

[Pharo-project] About UITheme, Widgets and coupling

by StÃ©phane Ducasse

Hi guys we got rid of preferences as a ugly facade and now we have UITheme and UIManager. I would like to understand how we can skin widgets while be as loosely coupled as possible. May be a UI builder is missingâ¦. I was looking at SystemProgressMorph we setDefaultParameters self theme setSystemProgressMorphDefaultParameters: self updateColor "update the apareance of aMorph" self theme preferGradientFill ifFalse: [^ self]. self fillStyle: (self theme progressFillStyleFor: self) and SystemProgressBarMorph has initialize "Initialize the receiver from the current theme." super initialize. barSize := 0. self fillStyle: ( self theme progressBarFillStyleFor: self); borderStyle: (self theme progressBarBorderStyleFor: self); barFillStyle: ( self theme progressBarProgressFillStyleFor: self) extent: aPoint "Update the bar fillStyle if appropriate." super extent: aPoint. self fillStyle isOrientedFill ifTrue: [ self fillStyle: (self theme progressBarFillStyleFor: self)]. self barFillStyle isOrientedFill ifTrue: [ self barFillStyle: ( self theme progressBarProgressFillStyleFor: self)] So to help me understanding the problem, I thought about the flow of things. I would really to avoid to have - widgets construction code in UITheme and UImanager (ref to debugger UI as mentioned by henryk) - UITheme invocation inside the widgets. May be a widget should have its own state and when we build it the class creation only push the info from the UITheme to the widgets What do you think? Stef

Dec. 6, 2011

Re: [Pharo-project] Squeak/Pharo in Fedora

by Pavel Krivanek

On Tue, Dec 6, 2011 at 9:58 PM, Eliot Miranda <eliot.miranda(a)gmail.com> wrote: > > > On Tue, Dec 6, 2011 at 12:56 PM, Pavel Krivanek <pavel.krivanek(a)gmail.com> > wrote: >> >> Hi Eliot, >> >> On Tue, Dec 6, 2011 at 9:38 PM, Eliot Miranda <eliot.miranda(a)gmail.com> >> wrote: >> > >> > >> > On Tue, Dec 6, 2011 at 11:05 AM, Pavel Krivanek >> > <pavel.krivanek(a)gmail.com> >> > wrote: >> >> >> >> Hi, >> >> >> >> I want make a little report about progress with making of Fedora >> >> pakcages for Smalltalk related projects (StackVM, CogVM, Pharo...) by >> >> Jaroslav Å karvada. >> >> >> >> Currently there are three packages in Fedora repositories - squeak-vm, >> >> squeak-image and squeak-vm-nonXOplugins (that contains plugins like >> >> OSProcess). It uses resources from the Squeak site. >> >> >> >> Red Hat has quite restrictive demands on packages and it is a small >> >> miracle that they tolerate current squeak packages :-). The conditions >> >> simply do expect that a program is a set of source codes and not a >> >> system of communicating objects. One problem is that Squeak VM is one >> >> of few metacircular packages that are needed for their own building. >> >> Because binary executables are prohibited, it would require a special >> >> bureaucratically problematic exception. Fortunately for virtual >> >> machine generation we may use the current squeak packages - more >> >> precisely their updated versions. The next issue for virtual machine >> >> generation may be the fact, that some packages used for VM generation >> >> are in repositories with public read&write permissions. >> >> >> >> Because image can be likened to *.jar files that are prohibited and >> >> must be recompiled during package generation, we are lucky that >> >> squeak-image package is currently accepted. If someone starts >> >> tinkering, we are in big trouble. >> >> >> >> There is a little issue with *.sources placement. The current VM >> >> package contains symlinks from VM path to sources files that are >> >> present in package with the image. This is really ugly solution. >> >> Squeak and Pharo uses this sequence for sources lookup: vm path, image >> >> path, current path. So I think that the best solution will be to have >> >> sources file in the image package, do not have symlinks in vm package >> >> and to create symlink to sources together with changes in the image >> >> path via mysqueak initialization script. >> >> >> >> Btw., Jaroslav told me that he saw a lot of creepy projects but >> >> nothing like Squeak VM :-) >> >> >> >> And there is a mess in VM versioning. What is the latest CogVM/StackVM >> >> version number? >> > >> > >> > How so? Â My latest version is SVN >> > http://www.squeakvm.org/svn/squeak/branches/Cog/ r2519. Â Monticello >> > package >> > http://source.squeak.org/VMMaker/VMMaker.oscog-eem.139. >> >> What is the relation between this SVN repository and Git repository >> used by the CI server (https://git.gitorious.org/cogvm/blessed.git)? > > > Not anything to do with me. It is in hands of Esteban. Strange, I thought that the Cog development is not splitted. The Git version uses cmake. SVN version autotools? >> What is the reason to exclude fbdev? > > > What's fbdev? framebuffer display device. >> >> >> -- Pavel >> >> >> Cheers, >> >> -- Pavel >> >> >> > >> > >> > >> > -- >> > best, >> > Eliot >> > >> > > > > -- > best, > Eliot >

Dec. 6, 2011

Re: [Pharo-project] Squeak/Pharo in Fedora

by Mariano Martinez Peck

How so? My latest version is SVN >>> http://www.squeakvm.org/svn/squeak/branches/Cog/ r2519. Monticello >>> package http://source.squeak.org/VMMaker/VMMaker.oscog-eem.139. >>> >> >> But where is such information stored? >> Then, the VMMaker package version says nothing, because to buiyld the VM >> you need more packages than VMMaker, so you have to guess which version to >> use. >> Moreoever, there is also ANOTHER version in VM which is VMMaker >> class>>versionString which is different from that. >> Then from the image or from a log we get: VM: Mac OS - intel - 1068 - >> Croquet Closure Cog VM [CoInterpreter VMMaker.oscog-eem.78] Croquet Cog >> 3.0.0 >> so where it is the svn version? >> > > (1007 to: 1009) collect: [:n| Smalltalk getSystemAttribute: n] > > #('CoInterpreter VMMaker.oscog-eem.138 uuid: > f9a63bdf-7bbc-4ae5-9634-ecd6fd814c9d Nov 18 2011' > 'StackToRegisterMappingCogit VMMaker.oscog-eem.139 uuid: > c2849383-9768-4948-b9b2-a5c22d482b07 Nov 18 2011' > 'r2519 http://www.squeakvm.org/svn/squeak/branches/Cog') > > Good enough? > Better. So I guess that at least in Pharo this method deserves an update: version "Return the version number string of the Virtual machine built" " Smalltalk vm version " ^(self getSystemAttribute: 1004), ' ', ((self getSystemAttribute: 1009) ifNil: ['']) In fact, only the last is required since the VM and plugin source is > included in http://www.squeakvm.org/svn/squeak/branches/Cog/src. > > >> So...all in all, I think that indeed, it is quite a mess. But it is also >> complicated to manage and I don't really know what to suggest. >> With ConfigurationOfCog at least with one number we group all image side >> related packages. >> Then we only need svn/git version. Moreover, we can store this svn/git as >> the description in each version of ConfigurationOfCog. That way, you know >> *exactly* what do you need. >> >> Cheers >> >> >> >>> >>> >>>> >>>> Cheers, >>>> -- Pavel >>>> >>>> >>> >>> >>> -- >>> best, >>> Eliot >>> >>> >> >> >> -- >> Mariano >> http://marianopeck.wordpress.com >> >> > > > -- > best, > Eliot > > -- Mariano http://marianopeck.wordpress.com

Dec. 6, 2011