[BUG] 0004990: #requiredSelectors is wrong

> http://bugs.impara.de/view.php?id=4990
>
> It seems there is a bug in 3.9.
> Here is a test case:
>
> ATestCase>>testRequirement
>   "
>   self debug: #testRequirement
>   "
>   | class |
>   class := Object
>             subclass: #AClassForTest
>             instanceVariableNames: ''
>             classVariableNames: ''
>             poolDictionaries: ''
>             category: self class category.
>   [
>    class compile: 'call
>                     ^ self isCalled'.
>    self assert: (class requiredSelectors includes: #isCalled).
>
>
>    class compile: 'isCalled
>                     ^ 1'.
>    "Fail here:"
>    self deny: (class requiredSelectors includes: #isCalled)]
>
>   ensure: [class removeFromSystem]
>

> Dear 3.9 release team,
>
> I realize that to be integrated at this late stage, any changes  
> going into the image should be of minimal impact, and of some  
> importance. I still advocate the integration of Traits-dvf.225.mcz  
> available at http://bugs.impara.de/view.php?id=4990 because:
> 1. It is based on the latest gamma release.
> 2. It is a one liner.
> 3. I am the author of the original code, and I can tell you it is a  
> quite conservative change (clear a cache when its needed).
> 4. The bug being fixed is covered by a test (now green) and so is  
> the rest of that code.
> 5. The code fixed "aClass requiredSelectors" has zero clients in  
> the image (so they won't regress) but hopefully, Damien will  
> provide a widespread client for it before 3.10 is available.  
> Therefore the fix is important.
>
> So how do we take the last step of making this actually go into the  
> image? do I reassign the bug to someone? I couldn't find this  
> documented on the wiki or mantis docs.
>
> Daniel
>
> Damien Cassou wrote:
>> Daniel Vainsencher a écrit :
>>> Thanks for testing! and for showing the bug in a very  
>>> reproducible manner. I've updated the mantis entry, with a  
>>> version that fixes the bug, please verify that that version works  
>>> for you (read the note).
>>
>> What will you do with this correction ? Can you publish it  
>> somewhere so that squeak-dev image can use it automatically ?
>>
>
>

> Dear 3.9 release team,
>
> I realize that to be integrated at this late stage, any changes  
> going into the image should be of minimal impact, and of some  
> importance. I still advocate the integration of Traits-dvf.225.mcz  
> available at http://bugs.impara.de/view.php?id=4990 because:
> 1. It is based on the latest gamma release.
> 2. It is a one liner.
> 3. I am the author of the original code, and I can tell you it is a  
> quite conservative change (clear a cache when its needed).
> 4. The bug being fixed is covered by a test (now green) and so is  
> the rest of that code.
> 5. The code fixed "aClass requiredSelectors" has zero clients in  
> the image (so they won't regress) but hopefully, Damien will  
> provide a widespread client for it before 3.10 is available.  
> Therefore the fix is important.
>
> So how do we take the last step of making this actually go into the  
> image? do I reassign the bug to someone? I couldn't find this  
> documented on the wiki or mantis docs.
>
> Daniel
>
> Damien Cassou wrote:
>> Daniel Vainsencher a écrit :
>>> Thanks for testing! and for showing the bug in a very  
>>> reproducible manner. I've updated the mantis entry, with a  
>>> version that fixes the bug, please verify that that version works  
>>> for you (read the note).
>>
>> What will you do with this correction ? Can you publish it  
>> somewhere so that squeak-dev image can use it automatically ?
>>
>
>

> Dear 3.9 release team,
>
> I realize that to be integrated at this late stage, any changes  
> going into the image should be of minimal impact, and of some  
> importance. I still advocate the integration of Traits-dvf.225.mcz  
> available at http://bugs.impara.de/view.php?id=4990 because:
> 1. It is based on the latest gamma release.
> 2. It is a one liner.
> 3. I am the author of the original code, and I can tell you it is a  
> quite conservative change (clear a cache when its needed).
> 4. The bug being fixed is covered by a test (now green) and so is  
> the rest of that code.
> 5. The code fixed "aClass requiredSelectors" has zero clients in  
> the image (so they won't regress) but hopefully, Damien will  
> provide a widespread client for it before 3.10 is available.  
> Therefore the fix is important.
>
> So how do we take the last step of making this actually go into the  
> image? do I reassign the bug to someone? I couldn't find this  
> documented on the wiki or mantis docs.
>
> Daniel
>
> Damien Cassou wrote:
>> Daniel Vainsencher a écrit :
>>> Thanks for testing! and for showing the bug in a very  
>>> reproducible manner. I've updated the mantis entry, with a  
>>> version that fixes the bug, please verify that that version works  
>>> for you (read the note).
>>
>> What will you do with this correction ? Can you publish it  
>> somewhere so that squeak-dev image can use it automatically ?
>>
>
>