On 2013-02-13, at 22:45, Christophe Demarey <Christophe.Demarey@inria.fr> wrote:

Hi,

Pharo 2.0 is shipped with Metacello. Metacello version installed is based on the stable version (1.0-beta.31.1.5). This version loads Metacello-ToolBox-dkh.130 but strangely, the Metacello-ToolBox package shipped in Pharo 2.0 is Metacello-ToolBox-MarkusDenker.135. Its ancestor is Metacello-ToolBox-dkh131 but Metacello-ToolBox-dkh.130 is not its ancestor. It seems that Metacello-ToolBox-dkh131 is another branch and that Pharo 2.0 does not ship the right version of Metacello Toolbox.

I found this with tests using the toolbox. They failed because the method configurationNameFrom behavior is not the same : Metacello-ToolBox-dkh.130>>Metacello-ToolBox>>configurationNameFrom: baseName "Return the fully-qualified configuration class name."

<apiDocumentation> ^ (baseName indexOfSubCollection: 'ConfigurationOf') > 0 ifTrue: [ baseName ] ifFalse: [ 'ConfigurationOf' , baseName ]

Metacello-ToolBox-MarkusDenker.135>>Metacello-ToolBox>>configurationNameFrom: baseName "Return the fully-qualified configuration class name."

<apiDocumentation> self flag: 'More work needed based on MetacelloScriptEngine'. ^ baseName' => Here it is strange because the basename of a Configuration is the name without 'ConfigurationOf' and e expect to have as result 'ConfigurationOfXXX'

So, 2 questions : is the Metacello-ToolBox-MarkusDenker.135 working?

nope, it only works partially. I already made some hackish fixes here and there to pass at least the #validate method. That's why I guess we are at 135.

should we not use the same base version of Metacello-ToolBox as the one declared (Metacello-ToolBox-dkh.130) in Metacello 1.0-beta.31.1.5 configuration?

I guess we should fix the current Toolbox by loading in the changes from 130. Initially we must have loaded an older version of the Toolbox and then start to fix certain things on it.