On Mon, Apr 29, 2013 at 11:08 AM, <commits(a)source.squeak.org> wrote:
> Colin Putney uploaded a new version of Traits to project The Trunk:
> http://source.squeak.org/trunk/Traits-cwp.293.mcz
>
> ==================== Summary ====================
>
> Name: Traits-cwp.293
> Author: cwp
> Time: 29 April 2013, 11:07:42.535 am
> UUID: f3c6cade-2e61-41c9-904d-9c77f00030a8
> Ancestors: Traits-cwp.292
>
> When registering a Trait, catch and resume AttemptToWriteReadOnlyGlobal,
> since there may already be a class binding.
>
> =============== Diff against Traits-cwp.292 ===============
>
This ought to fix the traits tests.
Do we need to put out a new alpha image with a recompileAll?
On Sun, Apr 28, 2013 at 12:44 PM, <commits(a)source.squeak.org> wrote:
> Colin Putney uploaded a new version of System to project The Trunk:
> http://source.squeak.org/trunk/System-cwp.526.mcz
>
> ==================== Summary ====================
>
> Name: System-cwp.526
> Author: cwp
> Time: 28 April 2013, 10:43:06.298 am
> UUID: d0a7abd6-1556-4900-b5ca-640d1aed4d7c
> Ancestors: System-ul.525
>
> Update Behavior>>allCallsOn to search for references using #bindingOf: rather than #associationAt:. This matches the binding created by the compiler.
>
> =============== Diff against System-ul.525 ===============
>
> Item was changed:
> ----- Method: Behavior>>allCallsOn (in category '*System-Support') -----
> allCallsOn
> "Answer a SortedCollection of all the methods that refer to me by name or as part of an association in a global dict."
>
> | theClass |
> theClass := self theNonMetaClass.
> + ^(self systemNavigation allCallsOn: (self environment bindingOf: theClass name)),
> - ^(self systemNavigation allCallsOn: (self environment associationAt: theClass name)),
> (Preferences thoroughSenders
> ifTrue: [OrderedCollection new]
> ifFalse: [self systemNavigation allCallsOn: theClass name]),
> (self systemNavigation allClasses
> select: [:c| c sharedPools includes: theClass]
> thenCollect:
> [:c|
> ClassReference new
> setClassSymbol: c name
> classIsMeta: false
> stringVersion: c name])!
>
>
On 29 April 2013 07:58, <commits(a)source.squeak.org> wrote:
> Colin Putney uploaded a new version of Environments to project The Trunk:
> http://source.squeak.org/trunk/Environments-cwp.25.mcz
>
> ==================== Summary ====================
>
> Name: Environments-cwp.25
> Author: cwp
> Time: 28 April 2013, 11:56:36.28 pm
> UUID: 0117c0ea-d81e-4eff-a6d6-afd791496d6d
> Ancestors: Environments-cwp.24
>
> Use aliases for classes imported with a different name, rather than copying the class binding.
>
> =============== Diff against Environments-cwp.24 ===============
This or its predecessor break the Traits tests. For instance:
http://build.squeak.org/job/SqueakTrunk/286/testReport/junit/TraitsTests.Ke…
I suspect it's just a case of making something Trait-y use #bindingOf:
or something? The tests seem to all fail for the same reason.
frank
Well, I think the proper change may be to make all implementors of
#versionNamesForPackageNamed: consistently returning the same thing.
It looks like I originally made this method answer an Array since 2011
but, only when I recently added it to MCRepositoryGroup, made THAT ONE
answer a Set. I failed to look at the others but instead "just
thought about it". If the same version was in more than one
repository we don't really need it multiple times.
So what is the correct behavior? Perhaps answering a Set is the
correct behavior afterall and the others should be changed too?
On Mon, Apr 29, 2013 at 6:13 AM, <commits(a)source.squeak.org> wrote:
> Frank Shearar uploaded a new version of Installer-Core to project The Trunk:
> http://source.squeak.org/trunk/Installer-Core-fbs.366.mcz
>
> ==================== Summary ====================
>
> Name: Installer-Core-fbs.366
> Author: fbs
> Time: 29 April 2013, 12:13:46.368 pm
> UUID: a93fa0fc-2a93-405a-86df-7aa334ee214b
> Ancestors: Installer-Core-fbs.365
>
> versionNames is a Set nowadays, but we want to sort it. Hence, convert to an Array.
>
> =============== Diff against Installer-Core-fbs.365 ===============
>
> Item was changed:
> ----- Method: InstallerMonticello>>mcThing (in category 'monticello') -----
> mcThing
> | loader |
> loader := self classMCVersionLoader new.
>
> "several attempts to read files - repository readableFileNames
> sometimes fails"
> self packages
> do: [:pkg |
> | versionNames fileToLoad version |
> versionNames := mc versionNamesForPackageNamed:
> (pkg asMCVersionName versionNumber = 0
> ifTrue: [ "Just a package name specified, use it whole." pkg ]
> ifFalse: [pkg asMCVersionName packageName]).
> + versionNames asArray sort: self mcSortFileBlock.
> - versionNames sort: self mcSortFileBlock.
> fileToLoad := versionNames
> detect: (self mcDetectFileBlock: pkg)
> ifNone: [ nil ].
> fileToLoad
> ifNotNil: [version := mc versionNamed: fileToLoad.
> (version isKindOf: MCConfiguration)
> ifTrue: [^ version]
> ifFalse: [self normalizedRepositories do: [:repo |
> MCRepositoryGroup default addRepository: repo].
> self normalizedRepositories do: [:repo |
> version workingCopy repositoryGroup addRepository: repo].
> loader addVersion: version].
> self logCR: ' found ' , version fileName , '...']].
> ^ loader!
>
>
On 29 April 2013 12:13, <commits(a)source.squeak.org> wrote:
> Frank Shearar uploaded a new version of Installer-Core to project The Trunk:
> http://source.squeak.org/trunk/Installer-Core-fbs.366.mcz
>
> ==================== Summary ====================
>
> Name: Installer-Core-fbs.366
> Author: fbs
> Time: 29 April 2013, 12:13:46.368 pm
> UUID: a93fa0fc-2a93-405a-86df-7aa334ee214b
> Ancestors: Installer-Core-fbs.365
>
> versionNames is a Set nowadays, but we want to sort it. Hence, convert to an Array.
>
> =============== Diff against Installer-Core-fbs.365 ===============
This will fix the ReleaseSqueakTrunk build. It _should_ have bailed
(because it should have been using CommandLineToolSet). I'll find out
why it didn't, and make it do so.
frank