- Pharo-dev - lists.pharo.org

[Pharo-project] [update] #10399

by Marcus Denker

10399 ----- Issue 1004: Pharo includes 4 classes named A, B, C, and D Issue 1002: harvester menu Issue 632: Intention Revealing: BasicClassOrganizer>>hasNoComment -- Marcus Denker - http://marcusdenker.de PLEIAD Lab - Computer Science Department (DCC) - University of Chile

July 29, 2009

Re: [Pharo-project] Beginner Tasks

by Simon Denier

Stef is talking of http://www.squeaksource.com/TestServerSimple.html Now this is a rather rudimentary adaptation from a tool made by Swiss guys. It runs automatically when launching the image, loads packages and runs tests for different projects. Right now it is tailored for Moose projects I guess that the area that need improvements are: - configuration of projects to load, with prerequisites (using Metacello perhaps?). Right now this is hard-coded, see TSProject class>>run - headless running (sometimes some interactive dialogs still pop up, especially when there is a problem loading a package, and this ruins the automatization) - better handling of errors (problems such as above) On 29 juil. 09, at 18:46, Robert Peters wrote: > > What is this "testServer" you are talking about? I searched this > forum a > little bit but didn't find any information on it. > > I write web-apps in Django for Google App Engine, and my current > test suite > for a project takes 45 seconds to run. This frustrates me because > unit > testing is (or, at least, should be) an easily distributed task (as > good > unit tests do not depend on each other). If Python would even > simply use > both cores of my main work computer, my tests would only take 23 > seconds, > and if it used all computers on my local network, it could take 5-10 > seconds. This would be a huge saving for me. > > All of this to say I have been thinking about a 'test server' that > would > take a queue of tests, set up the proper environments on as many > cores as it > had available, and send out the tests, collect the responses and > display > them. > > Is this what you mean by 'testServer'? > > I have wanted to help Pharo, but being an un-skilled Smalltalk > programmer, > and having other Python-based projects that were more urgent, I > haven't done > much about it. However, this 'testServer' caught my eye, because > I've been > thinking about creating such a thing for my own Python-based use. > Please > let me know if what I propose is what you intend. If so I may be > willing to > try and create something for Pharo to use. > > Kind regards, > Robert > > > StÃ©phane Ducasse-2 wrote: >> >> >> - a nicer project would be to help improving the embryonic >> testServer >> we have >> >> Stef >> >> > > -- > View this message in context: http://n2.nabble.com/Beginner-Tasks-tp3187343p3351068.html > Sent from the Pharo Smalltalk mailing list archive at Nabble.com. > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project -- Simon

July 29, 2009

Re: [Pharo-project] SmalltalkImage current platformName [WAS] Re: Baseline test results on pharo core 10392

by Andreas Wacknitz

Am 29.07.2009 um 21:13 schrieb Mariano Martinez Peck: > > > On Wed, Jul 29, 2009 at 6:06 PM, Andreas Wacknitz > <A.Wacknitz(a)gmx.de> wrote: > > Am 29.07.2009 um 20:08 schrieb Marcus Denker: > > > > > On 29.07.2009, at 14:02, Andreas Wacknitz wrote: > > > >> > >> Am 29.07.2009 um 17:46 schrieb Damien Cassou: > >> > >>> On Wed, Jul 29, 2009 at 5:20 PM, Andreas Wacknitz<A.Wacknitz(a)gmx.de > > > >>> wrote: > >>>> When I learned object orientation one of the first things I > >>>> encountered was: > >>>> "Procedural code makes decisions, object oriented code > delegates." > >>>> Or, sometimes simply: "Tell, don't ask!". > >>>> For me this looks like asking... > >>> > >>> I agree but this would be a good start I think. > >> A start in what direction? Is there a plan to change this later? > And > >> if so, when, who and how? > >> > > > > The old Squeak desease was that nothing ever was done, because at > some > > point > > it was found that "we could do better". This resulted in *complete* > > paralysis. > > > > For Pharo, the idea is to do something if itÂ´s better. Just imagine > > what you can > > do in 10 years if you do a lot of little things a little bit better. > > Instead of > > doing nothing and waiting for perfection. > > > I understand your point but part of Squeak's problem is its actual > state. > Many things are really messy and crappy. > While the proposed solution might be better for the actual problem (a > special unit test for the Mac), > I really fear a new source for code rot. When this isXXX will be > established it will be used in other situations > > Yes. I am agree. The thing is, who will do it? Ho has the time to > understand every place and change it ? > > > sooner or later. And I can't see that > OSPlatform current isMacOSX > is really that much better than > SmalltalkImage current platformName = 'Mac OS' > > > Yes it is. I it is not MUCH better, but a bit better and easy to > implement. Tomorrow the platform name changes and you have to change > your code all over the image. At least, with this, you have the > hardcoded in a single place. After this, with a single "senders" you > can see where it is used and how to modify this for a more OO stuff. I beg to disagree. First, you could also search for senders of "platformName". Second, what to do when you need to add a new platform? At the moment there is "isUnix". But there might be differences between Linux, Solaris, RiscOS, FreeBSD, ... You don't want a single search and multiple change but a single search AND change. In my opinion the proposed code is only a minor improvement. Andreas

July 29, 2009

Re: [Pharo-project] SmalltalkImage current platformName [WAS] Re: Baseline test results on pharo core 10392

by Mariano Martinez Peck

On Wed, Jul 29, 2009 at 6:06 PM, Andreas Wacknitz <A.Wacknitz(a)gmx.de> wrote: > > Am 29.07.2009 um 20:08 schrieb Marcus Denker: > > > > > On 29.07.2009, at 14:02, Andreas Wacknitz wrote: > > > >> > >> Am 29.07.2009 um 17:46 schrieb Damien Cassou: > >> > >>> On Wed, Jul 29, 2009 at 5:20 PM, Andreas Wacknitz<A.Wacknitz(a)gmx.de> > >>> wrote: > >>>> When I learned object orientation one of the first things I > >>>> encountered was: > >>>> "Procedural code makes decisions, object oriented code delegates." > >>>> Or, sometimes simply: "Tell, don't ask!". > >>>> For me this looks like asking... > >>> > >>> I agree but this would be a good start I think. > >> A start in what direction? Is there a plan to change this later? And > >> if so, when, who and how? > >> > > > > The old Squeak desease was that nothing ever was done, because at some > > point > > it was found that "we could do better". This resulted in *complete* > > paralysis. > > > > For Pharo, the idea is to do something if itÂ´s better. Just imagine > > what you can > > do in 10 years if you do a lot of little things a little bit better. > > Instead of > > doing nothing and waiting for perfection. > > > I understand your point but part of Squeak's problem is its actual > state. > Many things are really messy and crappy. > While the proposed solution might be better for the actual problem (a > special unit test for the Mac), > I really fear a new source for code rot. When this isXXX will be > established it will be used in other situations Yes. I am agree. The thing is, who will do it? Ho has the time to understand every place and change it ? > > sooner or later. And I can't see that > OSPlatform current isMacOSX > is really that much better than > SmalltalkImage current platformName = 'Mac OS' > Yes it is. I it is not MUCH better, but a bit better and easy to implement. Tomorrow the platform name changes and you have to change your code all over the image. At least, with this, you have the hardcoded in a single place. After this, with a single "senders" you can see where it is used and how to modify this for a more OO stuff. > > > > Andreas > > > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project >

July 29, 2009

Re: [Pharo-project] SmalltalkImage current platformName [WAS] Re: Baseline test results on pharo core 10392

by Andreas Wacknitz

Am 29.07.2009 um 20:08 schrieb Marcus Denker: > > On 29.07.2009, at 14:02, Andreas Wacknitz wrote: > >> >> Am 29.07.2009 um 17:46 schrieb Damien Cassou: >> >>> On Wed, Jul 29, 2009 at 5:20 PM, Andreas Wacknitz<A.Wacknitz(a)gmx.de> >>> wrote: >>>> When I learned object orientation one of the first things I >>>> encountered was: >>>> "Procedural code makes decisions, object oriented code delegates." >>>> Or, sometimes simply: "Tell, don't ask!". >>>> For me this looks like asking... >>> >>> I agree but this would be a good start I think. >> A start in what direction? Is there a plan to change this later? And >> if so, when, who and how? >> > > The old Squeak desease was that nothing ever was done, because at some > point > it was found that "we could do better". This resulted in *complete* > paralysis. > > For Pharo, the idea is to do something if itÂ´s better. Just imagine > what you can > do in 10 years if you do a lot of little things a little bit better. > Instead of > doing nothing and waiting for perfection. > I understand your point but part of Squeak's problem is its actual state. Many things are really messy and crappy. While the proposed solution might be better for the actual problem (a special unit test for the Mac), I really fear a new source for code rot. When this isXXX will be established it will be used in other situations sooner or later. And I can't see that OSPlatform current isMacOSX is really that much better than SmalltalkImage current platformName = 'Mac OS' Andreas

July 29, 2009

Re: [Pharo-project] SmalltalkImage current platformName [WAS] Re: Baseline test results on pharo core 10392

by Igor Stasenko

2009/7/29 Nicolas Cellier <nicolas.cellier.aka.nice(a)gmail.com>: > 2009/7/29 Igor Stasenko <siguctua(a)gmail.com>: >> 2009/7/29 Damien Cassou <damien.cassou(a)gmail.com>: >>> 2009/7/29 Mariano Martinez Peck <marianopeck(a)gmail.com>: >>>> >>>> On Tue, Jul 28, 2009 at 5:39 PM, Igor Stasenko <siguctua(a)gmail.com> wrote: >>>>> >>>>> I think that all such kind of queries (like isXXXPlatform) should be >>>>> not included and not used in a well-designed system. >>>>> If you want some capability - then you should ask directly for it, >>>>> irrespectively on what platform you are currently running. >>>> >>>> I am agree. I really believe in the phrase "tell don't ask". However, I have >>>> a couple of things to say: >>>> >>>> 1) It is not as easy as it seems. If you want to delegate to another object >>>> and be polymorphic regarding the platform you must reify a couple of classes >>>> (in your slash it is easy because you have the File hierarchy). I mean, you >>>> cannot put everything in the Platform class. And this must be done in every >>>> place. You must really understand that piece of code to change it. I can do >>>> this in my projects, but I don't know about to do it everywhere. >>> >>> The class OSPlatform is quite small. I would vote for: >>> >>> - implementing the following methods >>> >>> OSPlatform>>isMacOSX >>> Â ^ false >>> OSPlatform>>isWin32 >>> Â ^ false >>> OSPlatform>>isUnix >>> Â ^ false >>> >>> and then >>> >>> MacOSXPlatform>>isMacOSX >>> Â ^ true >>> UnixPlatform>>isUnix >>> Â ^ true >>> Win32Platform>>isWin32 >>> Â ^ true >>> >>> - when you want to test the platform, just do: >>> >>> OSPlatform current isMacOSX >>> >>> or >>> >>> OSPlatform current isWin32 >>> >>> What do you think? >>> >> >> How about shortcut methods then? >> >> OSPlatform class>>isMacOSX >> Â ^ self current isMacOSX >> >> etc... >> > > Or let it be a global? like Smalltalk, Processor, etc... > Do we write SystemDictionary current? > you mean global var 'Platform'? there is OSPlatform global var. Why not use it instead? Or, someone please tell me, why everyone tend to mis-use the singleton pattern in a form: 'Foo current' , whereas the Foo is already a singleton??? >>> -- >>> Damien Cassou >>> http://damiencassou.seasidehosting.st >>> >>> "Lambdas are relegated to relative obscurity until Java makes them >>> popular by not having them." James Iry >>> >>> _______________________________________________ >>> Pharo-project mailing list >>> Pharo-project(a)lists.gforge.inria.fr >>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project >>> >> >> >> >> -- >> Best regards, >> Igor Stasenko AKA sig. >> >> _______________________________________________ >> Pharo-project mailing list >> Pharo-project(a)lists.gforge.inria.fr >> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project > -- Best regards, Igor Stasenko AKA sig.

July 29, 2009

Re: [Pharo-project] SmalltalkImage current platformName [WAS] Re: Baseline test results on pharo core 10392

by Nicolas Cellier

2009/7/29 Igor Stasenko <siguctua(a)gmail.com>: > 2009/7/29 Damien Cassou <damien.cassou(a)gmail.com>: >> 2009/7/29 Mariano Martinez Peck <marianopeck(a)gmail.com>: >>> >>> On Tue, Jul 28, 2009 at 5:39 PM, Igor Stasenko <siguctua(a)gmail.com> wrote: >>>> >>>> I think that all such kind of queries (like isXXXPlatform) should be >>>> not included and not used in a well-designed system. >>>> If you want some capability - then you should ask directly for it, >>>> irrespectively on what platform you are currently running. >>> >>> I am agree. I really believe in the phrase "tell don't ask". However, I have >>> a couple of things to say: >>> >>> 1) It is not as easy as it seems. If you want to delegate to another object >>> and be polymorphic regarding the platform you must reify a couple of classes >>> (in your slash it is easy because you have the File hierarchy). I mean, you >>> cannot put everything in the Platform class. And this must be done in every >>> place. You must really understand that piece of code to change it. I can do >>> this in my projects, but I don't know about to do it everywhere. >> >> The class OSPlatform is quite small. I would vote for: >> >> - implementing the following methods >> >> OSPlatform>>isMacOSX >> Â ^ false >> OSPlatform>>isWin32 >> Â ^ false >> OSPlatform>>isUnix >> Â ^ false >> >> and then >> >> MacOSXPlatform>>isMacOSX >> Â ^ true >> UnixPlatform>>isUnix >> Â ^ true >> Win32Platform>>isWin32 >> Â ^ true >> >> - when you want to test the platform, just do: >> >> OSPlatform current isMacOSX >> >> or >> >> OSPlatform current isWin32 >> >> What do you think? >> > > How about shortcut methods then? > > OSPlatform class>>isMacOSX > Â ^ self current isMacOSX > > etc... > Or let it be a global? like Smalltalk, Processor, etc... Do we write SystemDictionary current? >> -- >> Damien Cassou >> http://damiencassou.seasidehosting.st >> >> "Lambdas are relegated to relative obscurity until Java makes them >> popular by not having them." James Iry >> >> _______________________________________________ >> Pharo-project mailing list >> Pharo-project(a)lists.gforge.inria.fr >> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project >> > > > > -- > Best regards, > Igor Stasenko AKA sig. > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

July 29, 2009

Re: [Pharo-project] SmalltalkImage current platformName [WAS] Re: Baseline test results on pharo core 10392

by Marcus Denker

On 29.07.2009, at 14:02, Andreas Wacknitz wrote: > > Am 29.07.2009 um 17:46 schrieb Damien Cassou: > >> On Wed, Jul 29, 2009 at 5:20 PM, Andreas Wacknitz<A.Wacknitz(a)gmx.de> >> wrote: >>> When I learned object orientation one of the first things I >>> encountered was: >>> "Procedural code makes decisions, object oriented code delegates." >>> Or, sometimes simply: "Tell, don't ask!". >>> For me this looks like asking... >> >> I agree but this would be a good start I think. > A start in what direction? Is there a plan to change this later? And > if so, when, who and how? > The old Squeak desease was that nothing ever was done, because at some point it was found that "we could do better". This resulted in *complete* paralysis. For Pharo, the idea is to do something if itÂ´s better. Just imagine what you can do in 10 years if you do a lot of little things a little bit better. Instead of doing nothing and waiting for perfection. Marcus

July 29, 2009

Re: [Pharo-project] SmalltalkImage current platformName [WAS] Re: Baseline test results on pharo core 10392

by Andreas Wacknitz

Am 29.07.2009 um 17:46 schrieb Damien Cassou: > On Wed, Jul 29, 2009 at 5:20 PM, Andreas Wacknitz<A.Wacknitz(a)gmx.de> > wrote: >> When I learned object orientation one of the first things I >> encountered was: >> "Procedural code makes decisions, object oriented code delegates." >> Or, sometimes simply: "Tell, don't ask!". >> For me this looks like asking... > > I agree but this would be a good start I think. A start in what direction? Is there a plan to change this later? And if so, when, who and how? Regards Andreas

July 29, 2009

Re: [Pharo-project] Beginner Tasks

by Robert Peters

What is this "testServer" you are talking about? I searched this forum a little bit but didn't find any information on it. I write web-apps in Django for Google App Engine, and my current test suite for a project takes 45 seconds to run. This frustrates me because unit testing is (or, at least, should be) an easily distributed task (as good unit tests do not depend on each other). If Python would even simply use both cores of my main work computer, my tests would only take 23 seconds, and if it used all computers on my local network, it could take 5-10 seconds. This would be a huge saving for me. All of this to say I have been thinking about a 'test server' that would take a queue of tests, set up the proper environments on as many cores as it had available, and send out the tests, collect the responses and display them. Is this what you mean by 'testServer'? I have wanted to help Pharo, but being an un-skilled Smalltalk programmer, and having other Python-based projects that were more urgent, I haven't done much about it. However, this 'testServer' caught my eye, because I've been thinking about creating such a thing for my own Python-based use. Please let me know if what I propose is what you intend. If so I may be willing to try and create something for Pharo to use. Kind regards, Robert StÃ©phane Ducasse-2 wrote: > > > - a nicer project would be to help improving the embryonic testServer > we have > > Stef > > -- View this message in context: http://n2.nabble.com/Beginner-Tasks-tp3187343p3351068.html Sent from the Pharo Smalltalk mailing list archive at Nabble.com.

July 29, 2009