On 24 Apr 2014, at 14:57, Camille Teruel <camille.ter...@gmail.com> wrote:

> 
> On 24 avr. 2014, at 14:37, Henrik Johansen <henrik.s.johan...@veloxit.no> 
> wrote:
> 
>> 
>> On 24 Apr 2014, at 2:21 , Marcus Denker <marcus.den...@inria.fr> wrote:
>> 
>>> 
>>> On 24 Apr 2014, at 14:04, Christophe Demarey <christophe.dema...@inria.fr> 
>>> wrote:
>>> 
>>>> Yes it looks like there is a bug in the slot class builder.
>>>> I also have a job failing since #30830
>>>> 
>>> 
>>> Yes, this is already here:
>>> 
>>>     
>>> https://pharo.fogbugz.com/f/cases/13223/Error-Should-not-happen-in-rebase-to
>>> 
>>> 
>>>     Marcus
>>> 
>> 
>> 
>> If I revert (comment out classModification propagate)
>> SlotClassBuilder >> #applyAndUpdateSharedVariableOrSharedPool: 
>> classModification
>>      ^ self track: classModification during: [ :old :new |
>>              installer classDefinitionChangedFrom: old to:  new by: 
>> classModification.
>>              classModification propagate ].
>> 
>> in a 80830 image, loading works fine.
>> 
>> … but I guess that change might be the entire point of the patch?
> 
> Yes removing this line will reintroduce the subclass bug. Moreover I think 
> this line is not the cause but a symptom revealer (cf my other mails about 
> classes that have wrong layout).
> Now I think the segfault is because of a wrong class format somewhere. 
> 

I revert this method in 3.0 834 so we get rid of the crashes… then we can find 
a better fix later.

        Marcus

Reply via email to