2014-08-02 10:18 GMT+02:00 Marcus Denker <marcus.denker@inria.fr>:

On 02 Aug 2014, at 08:54, stepharo <stepharo@free.fr> wrote:

> Yes I saw that too when we worked with Damien yesterday.
> May the last integration of clement was shaking some extensions and we get from time to time
> dirty packages that are not dirty in fact. It never occurred to me or marcus fixed them.
>

I think it is a side effect of the ��recompile class�� fix that now forced the trains to be re-installed.
This seems to emit notifications that it should not.

I added a OpalCompiler recompileAll as a postLoadDoIt in the last integration. I think this is what makes some packages dirty.��

This is because we recently patched recompileAll because it was not recompiling some methods in Traits.

��

> Stef
>
>
>> Loading and starting a fresh image
>> (Pharo 40148) shows some dirty package in monticello.
>> (KernelTests, CollectionsTests, ...)
>> Comparing those packages with pharo40inbox shows no
>> changes except for
>> NativeBoost-Core
>> Athens-Cairo
>> and AST-Core
>>
>> what happened?
>
>