October 2010 - Pharo-users - lists.pharo.org

Re: [Pharo-users] [Pharo-project] [ANN] Continue working in SimpleLogger
by StÃ©phane Ducasse Oct. 21, 2010

Oct. 21, 2010

Probably now we can cretae a pharo project and port it. The license is BSD. Stef On Oct 21, 2010, at 10:07 AM, Davorin Rusevljan wrote: > On Thu, Oct 21, 2010 at 10:03 AM, StÃ©phane Ducasse > <stephane.ducasse(a)inria.fr> wrote: >> come on tooth pick has even documentation and it does not look complex. >> redoing all over things does not reduce bugs and >> >> http://www.metaprog.com/Toothpick/docs/quickstart.html >> It looks clean and simple. > > Is squeak version working ok on Pharo? (I guess it does, but just checking) > > rush > http://www.cloud208.com/ > > _______________________________________________ > Pharo-users mailing list > Pharo-users(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-users

1 0

Re: [Pharo-users] [Pharo-project] [ANN] Continue working in SimpleLogger
by StÃ©phane Ducasse Oct. 21, 2010

Oct. 21, 2010

come on tooth pick has even documentation and it does not look complex. redoing all over things does not reduce bugs and http://www.metaprog.com/Toothpick/docs/quickstart.html It looks clean and simple. Stef On Oct 21, 2010, at 9:51 AM, Mariano Martinez Peck wrote: > > > On Thu, Oct 21, 2010 at 7:24 AM, StÃ©phane Ducasse <stephane.ducasse(a)inria.fr> wrote: > why don;t you look look at toothpick? > > > Because I don't have time to learn a new logging framework, nor to make it work on Pharo. I don't plan to be part of a logging framework. I just needed something really EASY for my projects. So, I remembered GermÃ¡n doing that, I know it was working in Pharo, and it took me 5 minutes to understand. 2 hours more to add my features, and that's all :) I just make it public for other users. > > Chees > > Mariano > > > On Oct 21, 2010, at 3:23 AM, Yanni Chiu wrote: > > > On 20/10/10 9:36 AM, Mariano Martinez Peck wrote: > >> What I did basically is to define SimpleLogger like the public API. But > >> then, this class delegates to a LoggerStrategy. I created the following > >> implementations: TranscriptLoggerStrategy, NullLoggerStrategy, > >> FileLoggerStrategy and FixesFileNameLoggerStrategy. In addition, I > >> created a superclass for them, marking the API with SimpleLogger. This > >> class is LoggerStrategy. > >> > >> So, now you can: > >> > >> - Enable and disable easily the logging (thanks to NullLoggerStrategy). > >> - Change on runtime WHERE or HOW to log (just change the strategy). > >> - Be able to write into a File, a Transcript, a FixedFileName, or nothing. > >> - Extend to you own strategy (FileAppendLoggerStrategy?? > >> SocketLoggerStrategy? etc...) > > > > My 2cents. I've not used or looked at SimpleLogger, but recall that it was a single class, plus a metacello config class. > > > > The new functionality sounds very useful, but is it still simple. It sounds like it was simple to add extra features - maybe because SimpleLogger was simple. Will the next person who comes along find a similarly simple implementation that can be easily adapted to specific needs. > > > > A hierarchy of strategy classes no longer sounds simple. What if I don't need all the different strategies loaded, will the configuration then allow multiple build targets for various individual or groups of strategies. Starting to sound complicated. > > > > IMHO, it should be a fork: SomewhatSimpleOnItsWayToComplicatedLogging :) > > > > -- > > Yanni > > > > > > _______________________________________________ > > Pharo-project mailing list > > Pharo-project(a)lists.gforge.inria.fr > > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project > > > _______________________________________________ > Pharo-users mailing list > Pharo-users(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-users > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

2 1

Re: [Pharo-users] [Pharo-project] [ANN] Continue working in SimpleLogger
by StÃ©phane Ducasse Oct. 21, 2010

Oct. 21, 2010

why don;t you look look at toothpick? On Oct 21, 2010, at 3:23 AM, Yanni Chiu wrote: > On 20/10/10 9:36 AM, Mariano Martinez Peck wrote: >> What I did basically is to define SimpleLogger like the public API. But >> then, this class delegates to a LoggerStrategy. I created the following >> implementations: TranscriptLoggerStrategy, NullLoggerStrategy, >> FileLoggerStrategy and FixesFileNameLoggerStrategy. In addition, I >> created a superclass for them, marking the API with SimpleLogger. This >> class is LoggerStrategy. >> >> So, now you can: >> >> - Enable and disable easily the logging (thanks to NullLoggerStrategy). >> - Change on runtime WHERE or HOW to log (just change the strategy). >> - Be able to write into a File, a Transcript, a FixedFileName, or nothing. >> - Extend to you own strategy (FileAppendLoggerStrategy?? >> SocketLoggerStrategy? etc...) > > My 2cents. I've not used or looked at SimpleLogger, but recall that it was a single class, plus a metacello config class. > > The new functionality sounds very useful, but is it still simple. It sounds like it was simple to add extra features - maybe because SimpleLogger was simple. Will the next person who comes along find a similarly simple implementation that can be easily adapted to specific needs. > > A hierarchy of strategy classes no longer sounds simple. What if I don't need all the different strategies loaded, will the configuration then allow multiple build targets for various individual or groups of strategies. Starting to sound complicated. > > IMHO, it should be a fork: SomewhatSimpleOnItsWayToComplicatedLogging :) > > -- > Yanni > > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

2 1

Re: [Pharo-users] [Pharo-project] [ANN] Continue working in SimpleLogger
by Mariano Martinez Peck Oct. 21, 2010

Oct. 21, 2010

On Thu, Oct 21, 2010 at 3:22 AM, Yanni Chiu <yanni(a)rogers.com> wrote: > On 20/10/10 9:36 AM, Mariano Martinez Peck wrote: > >> What I did basically is to define SimpleLogger like the public API. But >> then, this class delegates to a LoggerStrategy. I created the following >> implementations: TranscriptLoggerStrategy, NullLoggerStrategy, >> FileLoggerStrategy and FixesFileNameLoggerStrategy. In addition, I >> created a superclass for them, marking the API with SimpleLogger. This >> class is LoggerStrategy. >> >> So, now you can: >> >> - Enable and disable easily the logging (thanks to NullLoggerStrategy). >> - Change on runtime WHERE or HOW to log (just change the strategy). >> - Be able to write into a File, a Transcript, a FixedFileName, or nothing. >> - Extend to you own strategy (FileAppendLoggerStrategy?? >> SocketLoggerStrategy? etc...) >> > > My 2cents. I've not used or looked at SimpleLogger, but recall that it was > a single class, plus a metacello config class. > > The new functionality sounds very useful, but is it still simple. It sounds > like it was simple to add extra features - maybe because SimpleLogger was > simple. Will the next person who comes along find a similarly simple > implementation that can be easily adapted to specific needs. > > A hierarchy of strategy classes no longer sounds simple. What if I don't > need all the different strategies loaded, will the configuration then allow > multiple build targets for various individual or groups of strategies. > Starting to sound complicated. > Yes, that's true. However, take a look at the code, and I promisde that in no more than 10 minutes you get the idea how to use it. There are 5 classes, with an average of 5 methods each. > > IMHO, it should be a fork: SomewhatSimpleOnItsWayToComplicatedLogging :) > > heheheh > -- > Yanni > > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project >

1 0

[ANN] Continue working in SimpleLogger
by Mariano Martinez Peck Oct. 20, 2010

Oct. 20, 2010

Hi guys (German). Today I needed to add some real log to something I am developing and I found that SimpleLogger missed those functionalities I was needing. So...I implemented them. It is 3 hours effort, so don't expect too much. But I just wanted to shared with you in case you need it. What I did basically is to define SimpleLogger like the public API. But then, this class delegates to a LoggerStrategy. I created the following implementations: TranscriptLoggerStrategy, NullLoggerStrategy, FileLoggerStrategy and FixesFileNameLoggerStrategy. In addition, I created a superclass for them, marking the API with SimpleLogger. This class is LoggerStrategy. So, now you can: - Enable and disable easily the logging (thanks to NullLoggerStrategy). - Change on runtime WHERE or HOW to log (just change the strategy). - Be able to write into a File, a Transcript, a FixedFileName, or nothing. - Extend to you own strategy (FileAppendLoggerStrategy?? SocketLoggerStrategy? etc...) I did these changes today morning so I didn't have time to write tests nor class comments. However, I wrote some class side example methods in SimpleLogger. Finally, for my own projects I want: - to use a FixedFileName (very similar to the way PharoDebug.log is used). - to only do "self log:" in my classes. The solution I did is that somewhere (maybe in a class of your project or maybe in a special class that holds this) in a class side variable I hold a Singleton with the Logger. So that you can always do for example: MyAppLogger logger log: 'zaraza'. And finally, I implemented Object >> log: (with category *MyAppPackage) log: anObject MyAppLogger logger log: anObject. With this, anywhere in the classes of my app I can do: self log: 'something'. The code is temporally commited in http://www.squeaksource.com/Marea German: if you think I should commit directly to SimpleLogger repo, just let me know. In addition, be completely free to change it as you want and commit it there too. Cheers mariano

1 0

Decompiling method body returns as block?
by HernÃ¡n Morales Durand Oct. 19, 2010

Oct. 19, 2010

Hi list, I'm using Pharo1.1rc2 #11400 and I want to extract the body part (the String without the selector) of a CompiledMethod, but I'm getting a BlockNode after sending #body and a MNU afterwards. This is what I do aCompiledMethod decompileWithTemps ---> a BytecodeAgnosticMethodNode which is displayed like: 'renderContentOn: aRenderer aRenderer form id: ''horizontalForm''; with: [aRenderer fieldSet: [ ... blablabla ]]' and is ok. Then aCompiledMethod decompileWithTemps body ---> a BlockNode displayed as: {[:aRenderer | aRenderer form id: 'horizontalForm'; with: [aRenderer fieldSet: [... blablabla]]]} I've expected the code string (sending #decompileString) without "[:aRenderer |" part (used to work that way). The thing is if I send #code to the BlockNode I get a "MessageNotUnderstood: CascadeNode>>code". I'd appreciate any hint to solve this. Cheers, -- HernÃ¡n Morales Information Technology Manager, Institute of Veterinary Genetics. National Scientific and Technical Research Council (CONICET). La Plata (1900), Buenos Aires, Argentina. Telephone: +54 (0221) 421-1799. Internal: 422 Fax: 425-7980 or 421-1799.

1 0

Re: [Pharo-users] Difference between Set and HashedCollection
by StÃ©phane Ducasse Oct. 18, 2010

Oct. 18, 2010

and we should clean it because it was introduced a bit fast in pharo :) On Oct 18, 2010, at 12:29 PM, Igor Stasenko wrote: > HashedCollection is an abstract base class, which contains common > behavior for all hash-based collections, > including Set, Dictionary etc. > > On 18 October 2010 13:04, Benjamin Van Ryseghem > <benjamin.vanryseghem(a)gmail.com> wrote: >> I was looking in Pharo sources and I do not understand the difference >> between Set and HashedCollection which both seems to work the same way. >> >> >> Benjamin >> _______________________________________________ >> Pharo-users mailing list >> Pharo-users(a)lists.gforge.inria.fr >> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-users >> >> > > > > -- > Best regards, > Igor Stasenko AKA sig. > > _______________________________________________ > Pharo-users mailing list > Pharo-users(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-users

2 1

Difference between Set and HashedCollection
by Benjamin Van Ryseghem Oct. 18, 2010

Oct. 18, 2010

I was looking in Pharo sources and I do not understand the difference between Set and HashedCollection which both seems to work the same way. Benjamin

2 1

FFI Problems in Pharo 1.1 'Callout mechanism not available'
by BraveSirRobin Oct. 17, 2010

Oct. 17, 2010

I'm having intermittent problems with FFI in Pharo 1.1. I often get my image to the point where I can only get the error 'Callout mechanism not available' whenever I call a DLL function. These functions have previously worked perfectly but then stop working for some reason. Anyone know what causes this and what the fix/workaround is ? The only solution I've found so far is to save my packages and load them into a new image which is a bit time consuming. BTW, I'm running under Windows 7, 32-bit. -- View this message in context: http://forum.world.st/FFI-Problems-in-Pharo-1-1-Callout-mechanism-not-avail… Sent from the Pharo Smalltalk Users mailing list archive at Nabble.com.

1 0

Oct. 12, 2010

==================== Summary ==================== Name: OpalCompiler-Core-MarcusDenker.47 Author: MarcusDenker Time: 7 October 2010, 10:10:56 am UUID: eabd712c-a974-4416-9551-e210d85b6949 Ancestors: OpalCompiler-Core-JorgeRessia.46 -> Pass over ASTTranslator to make the effect (after) vs. value visiting more correct. More tests are green. -> Cleanups ==================== Summary ==================== Name: OpalCompiler-Tests-MarcusDenker.48 Author: MarcusDenker Time: 12 October 2010, 10:14 am UUID: 53fb460f-935c-4dd7-87df-92b9c77eb1f4 Ancestors: OpalCompiler-Tests-MarcusDenker.47 for now, remove OCCompilationEquivalencyTest ==================== Summary ==================== Name: OpalCompiler-Core-MarcusDenker.48 Author: MarcusDenker Time: 12 October 2010, 10:15:08 am UUID: f26101a1-d417-47c7-ab02-4974f0a15c93 Ancestors: OpalCompiler-Core-MarcusDenker.47 - some simplifications (eg remove compiledMethod...Using: methods from bc builder and iRBuilder) - cleanups - some fixes/hacks to make more tests green. ==================== Summary ==================== Name: OpalCompiler-Core-MarcusDenker.49 Author: MarcusDenker Time: 12 October 2010, 4:06:20 pm UUID: 21e13b3b-418a-404c-9405-9438b939a7a7 Ancestors: OpalCompiler-Core-MarcusDenker.48 - ASTTranslator: now #acceptBlockNode is responsible for pushing nil on empty bloks - ASTTranslator: simplify #acceptSequenceNode ==================== Summary ==================== Name: OpalCompiler-Tests-MarcusDenker.49 Author: MarcusDenker Time: 12 October 2010, 4:09:30 pm UUID: 9a662177-0e87-49fc-9cc3-475a6678f4f5 Ancestors: OpalCompiler-Tests-MarcusDenker.48 - started to merge OCSpecialVariableCases with OCOpalCompilerExamples - removed all #expectedTestfailures (e.g. FFI related tests are just skipped when FFI not installed) - add ASTTranslator test for exampleToDoInsideTemp and testExampleToDoInsideTempNotInlined - added testPushArray and testTodoValue (and example methods) - test initialize class side of BytecodeGenerator

1 0