RecentMessages already handles display of missing methods, why do we
need purgeMissingMethods, et al anyway?

On Fri, Oct 4, 2013 at 10:16 AM, Frank Shearar <frank.shearar at gmail.com> wrote:
> These aren't testing methods, they're not special access that only
> tests need. They're cleanup functionality called by the user. The
> original methods in Utilities had _no_ senders (because they had no
> tests either).
>
> They're also not methods that cause an inter-package dependency in the
> wrong direction, so the reason for moving them (my reorganising work)
> doesn't apply here. Specifically, it doesn't matter that Tests depends
> on System, because it _must_ do so.
>
> frank
>
> On 4 October 2013 16:04, Chris Muller <asqueaker at gmail.com> wrote:
>> They were moved to Tests by the same basis you've been reorganizing
>> lots and lots of methods -- to the package where resides their only
>> senders.
>>
>> On Fri, Oct 4, 2013 at 2:48 AM, Frank Shearar <frank.shearar at gmail.com> wrote:
>>> On 03 Oct 2013, at 19:52, commits at source.squeak.org wrote:
>>>
>>>> Chris Muller uploaded a new version of Tests to project The Trunk:
>>>> http://source.squeak.org/trunk/Tests-cmm.256.mcz
>>>>
>>>> ==================== Summary ====================
>>>>
>>>> Name: Tests-cmm.256
>>>> Author: cmm
>>>> Time: 3 October 2013, 2:52:17.76 pm
>>>> UUID: c46a7f7b-159f-4f8c-9e26-ab6f97290614
>>>> Ancestors: Tests-nice.255
>>>>
>>>> Changes for updates to RecentMessages in System-cmm.600.
>>>>
>>>> =============== Diff against Tests-nice.255 ===============
>>>>
>>>> Item was added:
>>>> + ----- Method: RecentMessages>>purge: (in category '*tests') -----
>>>> + purge: aMethodReference
>>>> +    methodReferences remove: aMethodReference.!
>>>>
>>>> Item was added:
>>>> + ----- Method: RecentMessages>>purgeMissingMethods (in category '*tests') -----
>>>> + purgeMissingMethods
>>>> +    methodReferences := methodReferences select: [:mref | |cls|
>>>> +        cls := mref actualClass.
>>>> +        cls notNil
>>>> +            and: [cls isInMemory]
>>>> +            and: [mref selector == #Comment or: [(cls compiledMethodAt: mref selector ifAbsent: [nil]) notNil]]].
>>>
>>> OK so in my defence I'm on my phone, and the pinhole. IWW on my email prevented me from seeing this as a moving of methods. I still don't get it though. This was part of Utilities' version of this, not test scaffolding.
>>>
>>> frank
>>>
>>>> Item was changed:
>>>>  ----- Method: RecentMessagesTest>>testMaximumSubmissionCountCapsReferenceCount (in category 'testing') -----
>>>>  testMaximumSubmissionCountCapsReferenceCount
>>>>      rm maximumSubmissionCount: 2.
>>>>      rm recordSelector: #foo forClass: Utilities inEnvironment: Smalltalk globals.
>>>>      rm recordSelector: #bar forClass: Utilities inEnvironment: Smalltalk globals.
>>>>      rm recordSelector: #baz forClass: Utilities inEnvironment: Smalltalk globals.
>>>> +    self assert: #bar equals: rm leastRecent selector.!
>>>> -    self assert: #bar equals: rm oldest selector.!
>>>>
>>>> Item was changed:
>>>>  ----- Method: RecentMessagesTest>>testOldestReturnsOldestSubmission (in category 'testing') -----
>>>>  testOldestReturnsOldestSubmission
>>>> +    self assert: nil equals: rm leastRecent description: 'Return nil if no submissions yet'.
>>>> -    self assert: nil equals: rm oldest description: 'Return nil if no submissions yet'.
>>>>      rm recordSelector: #foo forClass: Utilities inEnvironment: Smalltalk globals.
>>>> +    self assert: #foo equals: rm leastRecent selector.
>>>> -    self assert: #foo equals: rm oldest selector.
>>>>      rm recordSelector: #baz forClass: Utilities inEnvironment: Smalltalk globals.
>>>> +    self assert: #foo equals: rm leastRecent selector.!
>>>> -    self assert: #foo equals: rm oldest selector.!
>>>>
>>>>
>>>
>>
>