+my own 2 cents on this one 2013/8/9 Igor Stasenko <siguctua@gmail.com>

my 2 cents. the method should be like that:

Behavior>>methodDict ^ methodDict

the tools should be aware that it is low-level accessor and result can be nil. Or use higher-level accessors provided by Class api (like #selectors etc) to avoid hacky coding.

On 9 August 2013 14:33, Clément Bera <bera.clement@gmail.com> wrote:

...

Hello Mariano and others,

In Pharo there is a method:

Behavior>>methodDict methodDict == nil ifTrue: [^ MethodDictionary new ]. ^ methodDict

It seems (according to the latest author) that it is Mariano who changed it from:

Behavior>>methodDict methodDict == nil ifTrue: [self error: 'MethodDict is nil']. ^ methodDict

A comment is present:

"The method dictionary of a class can be nil when we want to use the #cannotInterpret: hook. Indeed when a class dictionary is nil, the VM sends the message cannotInterpret: to the receiver but starting the look up in the superclass of the class whose method dictionary was nil. Now the system relies that when the message methodDict is sent to a class a method dictionary is returned. In order to prevent the complaints of tools and IDE unaware of this feature, we fool them by providing an empty MethodDictionary. This will hopefully work in most cases, but the tools will loose the ability to modify the behavior of this behavior. The user of #cannotInterpret: should be aware of this."

Now my problem is when I do:

MyClass methodDict: nil. MyClass new name.

It crashes the image in latest Pharo 3.

Mariano, do you have an idea on how to fix it ? I ask you because you are the latest author and it seems tat the bug comes from this method.

Thanks for any answer.

-- Best regards, Igor Stasenko.