On 10 May 2011 13:56, Mariano Martinez Peck <marianopeck@gmail.com> wrote:
On Tue, May 10, 2011 at 1:24 PM, Igor Stasenko <siguctua@gmail.com> wrote:
Okay, seems like i found the bug.
(CompiledMethodTrailer new sourcePointer: 0) testEncoding
fails.
And it happens only when you pass 0 as source pointer!! Such an obscure case!
So...this means that after a day and a half of work of the whole community we finally fix it?? I am happy. Thanks Igor and the rest who helped as well :)
Now...I am not sure if I understand. I saw your changes and you test. But my question is what happen with those incorrect CM that are arround. It means that the next build image, when it does the CompiledMethod cleanUp and destroy the sourcePointers, I will let correct CMs ? Does it mean that if I build a dev image now with such fix, I won't have these weird CMs bothering me? Ok....I am trying now....
Of course they will bother you, because they will hang around if there are closures referencing them
Thanks a lot to everybody for the effort,
Mariano
-- Best regards, Igor Stasenko AKA sig.