Squeak-dev August 2023

squeak-dev@lists.squeakfoundation.org

30 participants
163 discussions

Etoys
by Edwin Ancaer 10 Aug '23

10 Aug '23

Hello, does anyone know about an Etoys image for Smalltalk6? The one I got from the download i squeakland seems too old: [edwin@ottopedi ~/Smalltalk6]$ ./etoys.sh This interpreter (vers. 68021) cannot read image file (vers. 6502). Press CR to quit... Thanks a lot, Edwin Ancaer *;

5 10

[DOC REFRESH] HymnParrotTalk-v.3.9 documentation, hosted on GitBook
by Robert Withers 10 Aug '23

10 Aug '23

Hey, I finally made some time for Squeak, as all my other projects are moving forwards, coasting. I took the time to translate some of what I have written into the Hymn documentation. All advice welcome. > https://projects.callistohouse.org/parrottalk/ ••• stumpy rabbit ❤️‍🔥🐰

1 0

borderWidth Confusion
by Jim Rosenberg 09 Aug '23

09 Aug '23

I have artistic content, developed in Squeak, published as .image files, which I preserve and move to other images (e.g. newer versions of Squeak) by saving as .morph and loading the morphs back in new images. Lately I have been loading into Squeak 5.3 .morphs from Squeak 4.3 and getting very strange results where borderWidth shows as 1 in Squeak 5.3 but instance variable borderWidth is 0 -- in Squeak 4.3 borderWidth shows as 0, as intended. It's taken me a while to figure this out, but I think I have it nailed down. I can fix the problem easily enough in my own content, but whether Squeak should be "fixed" is not clear; I would hate for a "fix" to break somebody else's content. The "root problem" here seems to be that a morph's borderWidth is at least in some circumstances being stored in *two places*: 1. ("The obvious"): Morph instance variable borderWidth. 2. ("The obscure"): Morph instance variable extension instance variable otherProperties at: #borderStyle instance variable width. What happens when they disagree? As of Squeak 4.3, a morph with borderWidth 0 and borderStyle width 1 will show as if both are 0; as of Squeak 5.3 (and continuing to 6.0) such a morph will show as if both are 1. The code I am using to "fix" my own content looks like this: self allMorphs do: [ :m | m valueOfProperty: #borderStyle ifPresentDo: [ :s | (m borderWidth = s width) ifFalse: [ s width: m borderWidth ] ] ] I.e.: setting borderStyle width to be the same as borderWidth works for me, but I have no way of knowing whether it will break somebody else's stuff. In my opinion having the borderWidth in two places is not a good idea, but may be necessary for somebody's backward compatibility. Comments?

2 4

The Trunk: Kernel-mt.1518.mcz
by commits＠source.squeak.org 09 Aug '23

09 Aug '23

Marcel Taeumel uploaded a new version of Kernel to project The Trunk: http://source.squeak.org/trunk/Kernel-mt.1518.mcz ==================== Summary ==================== Name: Kernel-mt.1518 Author: mt Time: 9 August 2023, 3:24:15.725576 pm UUID: 8260fcb3-5465-1b43-a2bc-4d045c933398 Ancestors: Kernel-mt.1517 Fixes #hasInstVarRef, now that we have compiled blocks embeddded. =============== Diff against Kernel-mt.1517 =============== Item was added: + ----- Method: CompiledBlock>>hasInstVarRef (in category 'scanning') ----- + hasInstVarRef + + self flag: #discuss. "mt: Should we move up more of the scanning protocol from CompiledMethod to CompiledCode? See senders of #hasInstVarRef and implementor in Context." + ^ self method hasInstVarRef! Item was changed: ----- Method: CompiledMethod>>hasInstVarRef (in category 'scanning') ----- hasInstVarRef + "Answer whether the receiver references an instance variable." - "Answer whether the method references an instance variable." + | printer | - | scanner end printer | - - scanner := InstructionStream on: self. printer := InstVarRefLocator new. + self codeLiteralsDo: [:compiledCode | | scanner end | + scanner := InstructionStream on: compiledCode. + end := compiledCode endPC. + [scanner pc <= end] whileTrue: [ + (printer interpretNextInstructionUsing: scanner) ifTrue: [^true]]]. + ^ false! - end := self endPC. - - [scanner pc <= end] whileTrue: [ - (printer interpretNextInstructionUsing: scanner) ifTrue: [^true]. - ]. - ^false!

1 0

Squeak 6.0: Morphic-mt.2027.mcz
by commits＠source.squeak.org 09 Aug '23

09 Aug '23

Marcel Taeumel uploaded a new version of Morphic to project Squeak 6.0: http://source.squeak.org/squeak60/Morphic-mt.2027.mcz ==================== Summary ==================== Name: Morphic-mt.2027 Author: mt Time: 9 August 2023, 2:05:39.773983 pm UUID: 7f8f22c1-c6a7-8e48-9472-9ea12d13e59c Ancestors: Morphic-mt.2026 Fix for loading older .morph files where borderStyle was not used consistently. =============== Diff against Morphic-mt.2026 =============== Item was added: + ----- Method: BorderedMorph>>readDataFrom:size: (in category 'objects from disk') ----- + readDataFrom: aDataStream size: varsOnDisk + + | bc bw | + super readDataFrom: aDataStream size: varsOnDisk. + + "Update borderStyle for older versions of the receiver. Use temps because of side-effects." + bc := borderColor. bw := borderWidth. + self borderColor: bc. + self borderWidth: bw. + + self flag: #todo. "mt: If we were to remove those instVars at some point, we need to increase the receivers version and add a proper conversion method via SmartRefStream. Still, this might break compatbility with foreign subclasses that rely on inst-var access."!

1 0

The Trunk: Morphic-mt.2120.mcz
by commits＠source.squeak.org 09 Aug '23

09 Aug '23

Marcel Taeumel uploaded a new version of Morphic to project The Trunk: http://source.squeak.org/trunk/Morphic-mt.2120.mcz ==================== Summary ==================== Name: Morphic-mt.2120 Author: mt Time: 9 August 2023, 2:03:15.609576 pm UUID: 431e9bf2-0d4e-2b47-8cba-7d2a4ad2f422 Ancestors: Morphic-mt.2119 Fix for loading older .morph files where borderStyle was not used consistently. =============== Diff against Morphic-mt.2119 =============== Item was added: + ----- Method: BorderedMorph>>readDataFrom:size: (in category 'objects from disk') ----- + readDataFrom: aDataStream size: varsOnDisk + + | bc bw | + super readDataFrom: aDataStream size: varsOnDisk. + + "Update borderStyle for older versions of the receiver. Use temps because of side-effects." + bc := borderColor. bw := borderWidth. + self borderColor: bc. + self borderWidth: bw. + + self flag: #todo. "mt: If we were to remove those instVars at some point, we need to increase the receivers version and add a proper conversion method via SmartRefStream. Still, this might break compatbility with foreign subclasses that rely on inst-var access."!

1 0

The Trunk: Tools-mt.1224.mcz
by commits＠source.squeak.org 09 Aug '23

09 Aug '23

Marcel Taeumel uploaded a new version of Tools to project The Trunk: http://source.squeak.org/trunk/Tools-mt.1224.mcz ==================== Summary ==================== Name: Tools-mt.1224 Author: mt Time: 9 August 2023, 1:31:00.763576 pm UUID: 9d479465-5f01-254a-aac3-c50067595733 Ancestors: Tools-mt.1223 In TreeBrowser, fix issue where saving a method with some special category selected would set that special category. =============== Diff against Tools-mt.1223 =============== Item was changed: ----- Method: Browser>>compileMessage:notifying: (in category 'code pane') ----- compileMessage: aText notifying: aController "Compile the code that was accepted by the user, placing the compiled method into an appropriate message category. Return true if the compilation succeeded, else false." | fallBackCategoryName originalSelectorName result fallBackMethodName | + self hasSpecialMessageCategorySelected - self selectedMessageCategoryName = ClassOrganizer allCategory ifTrue: [ "User tried to save a method while the ALL category was selected" fallBackCategoryName := selectedMessageCategoryName. fallBackMethodName := selectedMessageName. editSelection == #newMessage ifTrue: [ "Select the 'as yet unclassified' category" selectedMessageCategoryName := nil. (result := self defineMessageFrom: aText notifying: aController) ifNil: ["Compilation failure: reselect the original category & method" selectedMessageCategoryName := fallBackCategoryName. selectedMessageName := fallBackMethodName] ifNotNil: [self setSelector: result]] ifFalse: [originalSelectorName := self selectedMessageName. self setOriginalCategoryIndexForCurrentMethod. selectedMessageName := fallBackMethodName := originalSelectorName. (result := self defineMessageFrom: aText notifying: aController) ifNotNil: [self setSelector: result] ifNil: [ "Compilation failure: reselect the original category & method" selectedMessageCategoryName := fallBackCategoryName. selectedMessageName := fallBackMethodName. ^ result notNil]]. self changed: #messageCategoryList. self changed: #messageList. ^ result notNil] ifFalse: [ "User tried to save a method while the ALL category was NOT selected" ^ (self defineMessageFrom: aText notifying: aController) notNil]! Item was added: + ----- Method: Browser>>hasSpecialMessageCategorySelected (in category 'message category list') ----- + hasSpecialMessageCategorySelected + ^ self selectedMessageCategoryName = ClassOrganizer allCategory! Item was added: + ----- Method: TreeBrowser>>hasSpecialMessageCategorySelected (in category 'message category list') ----- + hasSpecialMessageCategorySelected + + ^ (selectedMessageCategoryName = ClassOrganizer allCategory + or: [selectedMessageCategoryName = self class extensionCategory]) + or: [selectedMessageCategoryName = self class coreCategory]!

1 0

The Trunk: ToolBuilder-Morphic-mt.345.mcz
by commits＠source.squeak.org 09 Aug '23

09 Aug '23

Marcel Taeumel uploaded a new version of ToolBuilder-Morphic to project The Trunk: http://source.squeak.org/trunk/ToolBuilder-Morphic-mt.345.mcz ==================== Summary ==================== Name: ToolBuilder-Morphic-mt.345 Author: mt Time: 9 August 2023, 12:55:27.112576 pm UUID: 645352a9-d870-8448-9048-ebfdf68b96d5 Ancestors: ToolBuilder-Morphic-mt.344 For debug/browse action invocation in buttons (via halo), add support for message sends, not just block closures. Now we can debug the buttons in the FileList tools. =============== Diff against ToolBuilder-Morphic-mt.344 =============== Item was changed: ----- Method: PluggableButtonMorphPlus>>browseImplementationOfActionSelector (in category 'debug menu') ----- browseImplementationOfActionSelector + action ifNil: [^ super browseImplementationOfActionSelector]. + + action isMessageSend + ifFalse: ["closure" + action homeMethod browse] + ifTrue: ["messages send" + (action arguments ifNil: [#()]) ifNotEmpty: [:args | args explore]. + ToolSet browseMethod: (action receiver class lookupSelector: action selector)] + ! - action - ifNil: [super browseImplementationOfActionSelector] - ifNotNil: [action homeMethod browse].! Item was changed: ----- Method: PluggableButtonMorphPlus>>debugAction (in category 'debug menu') ----- debugAction action ifNil: [^ super debugAction]. (Process forBlock: [self performAction] + runUntil: [:context | context closure = action or: [context selector = action "message send" selector]]) - runUntil: [:context | context closure = action]) debugWithTitle: ('Debug button action "{1}" in model "{2}"' translated format: {self label. self target printString}).!

1 0

#withoutDuplicates on Collection?
by christoph.thiede＠student.hpi.uni-potsdam.de 08 Aug '23

08 Aug '23

Hi all, I was just wondering why we only define #withoutDuplicates on SequenceableCollection, since it does not depend on the order of the receiver and could be applied to other types of collections as well. For instance, Set could override it with ^self copy, Bags would be naturally covered as well, etc. What do you think? Best, Christoph --- Sent from Squeak Inbox Talk

9 17

Squeak 6.0: MorphicExtras-mt.320.mcz
by commits＠source.squeak.org 08 Aug '23

08 Aug '23

Marcel Taeumel uploaded a new version of MorphicExtras to project Squeak 6.0: http://source.squeak.org/squeak60/MorphicExtras-mt.320.mcz ==================== Summary ==================== Name: MorphicExtras-mt.320 Author: mt Time: 8 August 2023, 10:47:59.839546 am UUID: ce73ef05-d09d-6849-a1d7-486b8d720da3 Ancestors: MorphicExtras-mt.319 Fixes GH #80: 'preserver trash' preference ceased to work The preference continued to work but the trash book morph pages were layouted incorrectly, as layouting now requires a fullBounds computation. Setting the position does not suffice anymore. =============== Diff against MorphicExtras-mt.319 =============== Item was changed: ----- Method: BookMorph>>insertPageLabel:morphs: (in category 'insert and delete') ----- insertPageLabel: labelString morphs: morphList | m c labelAllowance | self insertPage. + labelString + ifNotNil: [ + m := labelString asMorph. + m + lock; + position: currentPage position + (((currentPage width - m width) // 2) @ 5). + currentPage addMorph: m. + labelAllowance := 40] + ifNil: [labelAllowance := 0]. - labelString ifNotNil: - [m := labelString asMorph. - m lock. - m position: currentPage position + (((currentPage width - m width) // 2) @ 5). - currentPage addMorph: m. - labelAllowance := 40] - ifNil: - [labelAllowance := 0]. "use a column to align the given morphs, then add them to the page" c := AlignmentMorph newColumn wrapCentering: #center; cellPositioning: #topCenter. c addAllMorphs: morphList. c position: currentPage position + (0 @ labelAllowance). + c fullBounds. "enforce layouting" currentPage addAllMorphs: morphList. ^ currentPage !

1 0

Jump to page: