On Sat, May 5, 2012 at 11:39 AM, Mariano Martinez Peck <marianopeck@gmail.com> wrote:

Good idea guys. I am trying now...

On Sat, May 5, 2012 at 9:43 AM, Marcus Denker <marcus.denker@inria.fr> wrote:

On May 4, 2012, at 11:46 PM, Guillermo Polito wrote:

>
> So...what do we do? �we just accept the depedency?
>

> Why not use a CompiledMethod instead? �More on, the TraitMethodDescription already knows the selector, so storing only the behaviors containing related methods should be enough...
>

Yes, the code is old from a time when CompiledMethod did not know the class or selector.

So this can be rewritten to just use compiledMethod.

So guys. I did an experiment with this. So I used a CompiledMethod rather than a RGMethodDefinition. I did a changeset (not sure if Monticello can handle these changes) just to have another one eye's on it. Basically, I needed to change the senders of #locatedMethods because now such method answers a collection of CompiledMethod rather than RGMethodDefinition.�

I tested with Fuel (fuel has lots of tests regarding traits) and even building PharoCore from PharoKernel without Ring in kernel and putting all packags in the same bundle.� (it works!!!).
However, the traits tests from Pharo are completly broken and it is impossible to do something. You run one test and you got 30 debuggers. It is impossible to test.

So, if someone can take a look, I really appreciate it.
http://code.google.com/p/pharo/issues/detail?id=5815

Thanks

�

� � � �Marcus

--
Marcus Denker -- http://marcusdenker.de

--
Mariano
http://marianopeck.wordpress.com