RefactoringManager and RBNamepsace error?

> Lukas, i have the following problem:
> model  := RBNamespace new.
> Then executing successive RenameRefactoring's using the same model yields an
> Error.
> 1. ClassA -> ClassA2 OK
> 2. ClassA2 -> ClassA3 OK
> 3. ClassA3 -> ClassA4  ERROR!
>
> I've tracked it down to this:
> Lukas, do you think i'm incorrectly using the namespace and refactoring's?

> 2010/3/19 Fernando olivero <[hidden email]>:
>> Lukas, i have the following problem:
>> model  := RBNamespace new.
>> Then executing successive RenameRefactoring's using the same model yields an
>> Error.
>> 1. ClassA -> ClassA2 OK
>> 2. ClassA2 -> ClassA3 OK
>> 3. ClassA3 -> ClassA4  ERROR!
>>
>> I've tracked it down to this:
>> Lukas, do you think i'm incorrectly using the namespace and refactoring's?
>
> Yeah, probably you are using it incorrectly.
>
> Just call #primtiveExecute on the refactoring if you only want to
> apply it to the RBNamespace model.
>
>>  Are namespace supposed to be used once and later ditched?
>
> RBNamespace provides a view on the system that can be manipulated
> without touching the actual system. The class has a horrible name, it
> should rather be called RBSmalltalkModel or something. After several
> operations all the changes can then be applied in a single "atomic"
> operation using:
>
>     aNamespace changes execute
>
>> My goal is to have a GauchoSystem which nows all the changes (with undos!)
>> that have happened, that's why i'm using always the same namespace.
>
> RBNamespace is a delta to the real system. You don't want to keep it
> around for too long. You perform some operations on it, you commit
> them to the image, and then you use a new instance. That's the basic
> workflow.
>
> The undoes are kept separate in RefactoryChangeManager.
>
> Lukas
>
> --
> Lukas Renggli
> http://www.lukas-renggli.ch
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

> Lukas, i have the following problem:
> model  := RBNamespace new.
> Then executing successive RenameRefactoring's using the same model yields an
> Error.
> 1. ClassA -> ClassA2 OK
> 2. ClassA2 -> ClassA3 OK
> 3. ClassA3 -> ClassA4  ERROR!
>
> I've tracked it down to this:
> 1.
> RenameRefactoring>>execute
> self primitiveExecute.
> RefactoringManager instance addRefactoring: self
> 2.
> RefactoringManager>>addRefactoring: aRefactoring
> RefactoryChangeManager instance performChange: aRefactoring changes.
> refactorings add: aRefactoring class name
> 3.
> Refactoring>>changes
> ^self model changes
> which answers all the changes!
> {  ClassA rename: #ClassA2!
>   ClassA2 rename: #ClassA3!
> ClassA3 rename: #ClassA4! }
> Why does a particular refactoring answers all the models changes? And not
> just the ones it introduced?
> The problem is in #performChange: will apply over and over all the changes,
> and not just the last one.
>
> Lukas, do you think i'm incorrectly using the namespace and refactoring's?
>  Are namespace supposed to be used once and later ditched?
> My goal is to have a GauchoSystem which nows all the changes (with undos!)
> that have happened, that's why i'm using always the same namespace.
>
> Thanks,
> Fernando
>
> pd: a test that reproduces the problem.
>
>>>testSucessiveRefactoringsError
> | namespace |
> namespace :=  RBNamespace new.
> [
> | refactoring|
> Object subclass: 'ClassA' asSymbol
> instanceVariableNames: ''
> classVariableNames: ''
> poolDictionaries: ''
> category: 'TestingNamespaces1234'.
> refactoring := RenameClassRefactoring
> model: namespace
> rename: (Smalltalk classNamed: 'ClassA' asSymbol )
> to: 'ClassA2' asSymbol .
> refactoring execute.
> refactoring := RenameClassRefactoring
> model: namespace
> rename:  (Smalltalk classNamed:'ClassA2' asSymbol )
> to:  'ClassA3' asSymbol.
> refactoring execute.
> refactoring := RenameClassRefactoring
> model: namespace
> rename:  (Smalltalk classNamed: 'ClassA3' asSymbol )
> to:  'ClassA4' asSymbol.
> self shouldnt:[refactoring execute] raise: Error description: 'Valid
> refactoring should not raise error'.
> ] ensure:[  SystemOrganization removeSystemCategory: 'TestingNamespaces1234'
> ]
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>