Rationale for RGClassDefinition>>definition?

is defined as
        ^ self realClass definition
instead of
        ^self definitionSource
and not two levels up at RGBehaviorDescription
to make it work for MetaClass and Traits.
Why is that?

Stephan

I don't know why it was defined like that. But for issue 17404 I changed that.

I do not know now I would have love that veronica believed in tests.
the difference is the following


a RG cna either
    - wrap and point to a living entity
    - acts a dead one (representing a class on a disk)

This difference is expressed in the RGMethod and probably not well in RGBehavior.
In fact for method I remember that there was another case.

But it has an impact.
In fact a RGEntity should have a strategy attach to it that defines how to get certain information
for such cases.

Ring should be revisited with such scenarios in mind.

Stef

I don't know why it was defined like that. But for issue 17404 I changed that.

Am 21.01.2016 12:15 vorm. schrieb "Stephan Eggermont" <[hidden email]>:

is defined as
        ^ self realClass definition
instead of
        ^self definitionSource
and not two levels up at RGBehaviorDescription
to make it work for MetaClass and Traits.
Why is that?

Stephan

016-01-21 9:07 GMT+01:00 stepharo <[hidden email]>:

I do not know now I would have love that veronica believed in tests.
the difference is the following


a RG cna either
    - wrap and point to a living entity
    - acts a dead one (representing a class on a disk)

Yes, I was about to add a check for isActive/isPassive for RGClassDefinition, but this isAcitve/isPassive is only

defined for Ringmethods not for Ring class (or behavior) definitions.

I think that we should add it to classes too.
In fact we should consider ring as a first try and it is in our plan to revisit it but too busy.