Squeak-dev December 2023

squeak-dev@lists.squeakfoundation.org

26 participants
124 discussions

Clean BlockClosures in Squeak
by christoph.thiede＠student.hpi.uni-potsdam.de 05 Dec '23

05 Dec '23

Hi all, this feels like a FAQ, but there have been so many similar decades-old posts on this list that I hope you forgive me for possibly re-posting this. :-) Simple example: Object subclass: #Dummy instanceVariableNames: '' classVariableNames: '' poolDictionaries: '' category: 'as yet unclassified' Dummy>>listen Smalltalk when: #plonk evaluate: [nil halt]. Dummy new listen. Smalltalk garbageCollect. self assert: Dummy allInstances isEmpty. "fail! our Dummy object is still referenced" chasePointers reveals that, as expected, our dummy is still captured inside the BlockClosure's receiver and outerContext. I have learned that I can use MessageSends and WeakMessageSends instead of a block closure in this particular purpose but am curious why clean blocks have to capture the receiver anyway. From as I understand FullBlockClosure comment, this would not be necessary, and indeed: Dummy>>listenClean Smalltalk when: #plonk evaluate: ([nil halt] outerContext: nil; receiver: nil; yourself). Dummy new listenClean. Smalltalk garbageCollect. self assert: Dummy allInstances = 1. "pass! just the previous reference" Smalltalk triggerEvent: #plonk. "still works" The only caveat is that this partially breaks BlockClosureInspector. So, long story short, could the compiler could indeed detect clean blocks and emit a special pushCleanBlockClosure instruction that does not unnecessarily reference the receiver? Or would there be any existing behavior that actually *relies on* keeping the receiver in the memory due to this? Do I assume correctly that clean blocks be possible and desired and are just not implemented at the moment? Asking further, could this even work for non-clean blocks (unless they have a non-local return or refer to self) thanks to the invention of closure vectors? This seems to work as well: Dummy>>listenClean2 | w | w := World. Smalltalk when: #plonk evaluate: ([w halt] outerContext: nil; receiver: nil; yourself). Asking because this can introduce severe memory leaks (if this is an appropriate term inside the image) and I think there is a lot of unnecessary metaprogramming with MessageSends and WeakMessageSends due to this in the Trunk. First, I would like to assure my understanding of the situation and second, I am dreaming of a future where blocks can rule all this. Plus, I'm very thankful there's no need for [](){} in Squeak. :-) Best, Christoph --- Sent from Squeak Inbox Talk

2 2

The Trunk: Morphic-mt.2147.mcz
by commits＠source.squeak.org 04 Dec '23

04 Dec '23

Marcel Taeumel uploaded a new version of Morphic to project The Trunk: http://source.squeak.org/trunk/Morphic-mt.2147.mcz ==================== Summary ==================== Name: Morphic-mt.2147 Author: mt Time: 4 December 2023, 11:09:57.559721 am UUID: 5c8a57d5-9661-924a-8ced-db7bf5d4664a Ancestors: Morphic-mt.2146 In object explorers, use a more compact representation for run-arrays. Thanks to Christoph (ct) for the idea! =============== Diff against Morphic-mt.2146 =============== Item was added: + ----- Method: RunArray>>explorerContents (in category '*Morphic-Explorer') ----- + explorerContents + "Instead of listing each element, choose a more compact representation like bags and sets do. Instead of #runsAndValuesDo:, choose an absolute representation of runs via #intervalsAndValuesDo:." + + ^ Array new: self runs size streamContents: [:stream | + self intervalsAndValuesDo: [:interval :value | + stream nextPut: (ObjectExplorerWrapper + with: value + name: interval printString + model: self)]]!

1 0

The Trunk: Morphic-mt.2146.mcz
by commits＠source.squeak.org 04 Dec '23

04 Dec '23

Marcel Taeumel uploaded a new version of Morphic to project The Trunk: http://source.squeak.org/trunk/Morphic-mt.2146.mcz ==================== Summary ==================== Name: Morphic-mt.2146 Author: mt Time: 4 December 2023, 10:41:12.878721 am UUID: 0270eb19-334f-bf46-9eea-dadbeeb695e7 Ancestors: Morphic-mt.2145 Complements Collections-mt.1057 for object explorers: - For bags, show number of occurances - For sets, drop that serial number to not imply index-based access =============== Diff against Morphic-mt.2145 =============== Item was added: + ----- Method: Bag>>explorerContents (in category '*Morphic-Explorer') ----- + explorerContents + "Show the receiver's contents in a compact way, similar to its #printString representation." + + ^ self valuesAndCounts withKeysSortedSafely associations withIndexCollect: [:assoc :index | + ObjectExplorerWrapper + with: assoc key + name: assoc value "i.e., number of occasions" printString, '×' + model: self]! Item was changed: ----- Method: Set>>explorerContents (in category '*Morphic-Explorer') ----- explorerContents + "Like dictionaries, indicate some human-readable sort order, yet refrain from showing a serial number to not imply index-based access." ^Array new: self size streamContents: [ :stream | + self sortedSafely do: [ :each | - self do: [ :each | stream nextPut: ( ObjectExplorerWrapper with: each + name: '·' "i.e., no order, no index-based access" - name: (stream position + 1) printString model: self) ] ]!

1 0

The Trunk: CollectionsTests-mt.402.mcz
by commits＠source.squeak.org 04 Dec '23

04 Dec '23

Marcel Taeumel uploaded a new version of CollectionsTests to project The Trunk: http://source.squeak.org/trunk/CollectionsTests-mt.402.mcz ==================== Summary ==================== Name: CollectionsTests-mt.402 Author: mt Time: 4 December 2023, 10:39:15.230721 am UUID: c8085214-d1bb-4a48-8c38-31a52ca02a4e Ancestors: CollectionsTests-ul.401 Complements Collections-mt.1057 =============== Diff against CollectionsTests-ul.401 =============== Item was added: + ----- Method: BagTest>>testPrintString (in category 'tests') ----- + testPrintString + + | bag | + bag := Bag new. + bag add: '1' withOccurrences: 5. + bag add: '2' withOccurrences: 1. + bag add: '3' withOccurrences: 3. + + self assert: 'a Bag(''1''->5 ''2''->1 ''3''->3)' equals: bag printString.! Item was added: + ----- Method: SetTest>>testPrintString (in category 'tests') ----- + testPrintString + + | set | + set := Set new. + set add: 'Squeak'. + set add: ','. + set add: 'Hello'. + + self assert: 'a Set('','' ''Hello'' ''Squeak'')' equals: set printString.!

1 0

The Trunk: Collections-mt.1057.mcz
by commits＠source.squeak.org 04 Dec '23

04 Dec '23

Marcel Taeumel uploaded a new version of Collections to project The Trunk: http://source.squeak.org/trunk/Collections-mt.1057.mcz ==================== Summary ==================== Name: Collections-mt.1057 Author: mt Time: 4 December 2023, 10:38:49.326721 am UUID: 3da620a5-90b9-f542-b1ab-6785eda55837 Ancestors: Collections-ct.1056 Update elements-printing for hashed collections, i.e., bags, sets, dictionaries: - For bags, choose the more compact way that dictionaries do - For sets, sort contents like dictionaries do for keys - For dictionaries, drop that trailing space Thanks to Christoph (ct) for the ideas! =============== Diff against Collections-ct.1056 =============== Item was added: + ----- Method: Bag>>printElementsOn: (in category 'printing') ----- + printElementsOn: aStream + "Pass request to internal contents dictionary to re-use its printString representation, which exposes (more compact) item counts and not a raw element list." + + contents printElementsOn: aStream.! Item was changed: ----- Method: Dictionary>>printElementsOn: (in category 'printing') ----- printElementsOn: aStream aStream nextPut: $(. self size > 100 ifTrue: [aStream nextPutAll: 'size '. self size printOn: aStream] ifFalse: [self keysInOrder do: [:key | aStream print: key; nextPutAll: '->'; + print: (self at: key)] + separatedBy: [aStream space]]. - print: (self at: key); - space]]. aStream nextPut: $)! Item was added: + ----- Method: Set>>printElementsOn: (in category 'printing') ----- + printElementsOn: aStream + "Like dictionaries, indicate some human-readable sort order." + + self sortedSafely printElementsOn: aStream.!

1 0

The Inbox: Tools-ct.1243.mcz
by commits＠source.squeak.org 04 Dec '23

04 Dec '23

Christoph Thiede uploaded a new version of Tools to project The Inbox: http://source.squeak.org/inbox/Tools-ct.1243.mcz ==================== Summary ==================== Name: Tools-ct.1243 Author: ct Time: 1 December 2023, 8:16:16.336752 pm UUID: 351663d4-0cc6-6d42-918a-e275ec86b5b0 Ancestors: Tools-mt.1241 Proposal: When accepting a method in a debugger, allows to proceed compilation despite of no active home method. I noticed a frequent demand for this in myself and others as in many situations, we are not interested in proceeding the suspended process anyway. The second dialog window is optional and might bring clarity or annoyance. Do we need it? =============== Diff against Tools-mt.1241 =============== Item was changed: ----- Method: Debugger>>contents:notifying: (in category 'accessing') ----- contents: aText notifying: aController "Accept new method source of the selected context." | selector classOfMethod category ctxt newMethod | contextStackIndex = 0 ifTrue: [^ false]. "First, handle some edge cases" selector := self selectedClass newParser parseSelector: aText. "selector isDoIt ifTrue: [ currentCompiledMethod := self compileDoIt: aText]." self flag: #todo. "ct: Recompile doIt method *without* creating method litters!! See Compiler>>#evaluateCue:ifFail:." selector = self selectedMessageName ifFalse: [ "Different message compiled, delegating to super" ^ super contents: aText notifying: aController]. self selectedContext isExecutingBlock ifTrue: [ "If we are in a block context, we need to rewind the stack before ." | home | home := self selectedContext activeHome. home ifNil: [ + (self confirm: 'Method for block not found on stack, can''t edit and continue. Edit anyway?' translated) ifFalse: [ + ^ false]. + ^ (super contents: aText notifying: aController) + ifTrue: [ + contents := nil. self contentsChanged. + self inform: 'Method was updated, but cannot revert the stack to it.' translated]; + yourself]. - self inform: 'Method for block not found on stack, can''t edit and continue' translated. - ^ false]. (self confirm: 'I will have to revert to the method from\which this block originated. Is that OK?' translated withCRs) ifFalse: [ ^ false]. self resetContext: home changeContents: false. "N.B. Only reset the contents if the compilation succeeds. If contents would be reset when compilation fails, both compiler error message and modifications were lost." ^ (self contents: aText notifying: aController) ifTrue: [self contentsChanged]; yourself]. classOfMethod := self selectedClass. category := self selectedMessageCategoryName. "Do the actual compilation" selector := classOfMethod compile: aText classified: category notifying: aController. selector ifNil: [^ false]. "compilation cancelled" "Update views" contents := aText. newMethod := classOfMethod compiledMethodAt: selector. newMethod isQuick ifTrue: [ self cutBackExecutionToSenderContext]. ctxt := interruptedProcess popTo: self selectedContext. ctxt == self selectedContext ifFalse: [self inform: 'Method saved, but current context unchanged\because of unwind error. Click OK to see error' translated withCRs] ifTrue: [ newMethod isQuick ifFalse: [ interruptedProcess restartTopWith: newMethod. interruptedProcess stepToSendOrReturn]. contextVariablesInspector object: nil]. self resetContext: ctxt. Project current addDeferredUIMessage: [ self changed: #contentsSelection]. ^ true!

2 1

The Inbox: Collections-ct.1058.mcz
by commits＠source.squeak.org 04 Dec '23

04 Dec '23

Christoph Thiede uploaded a new version of Collections to project The Inbox: http://source.squeak.org/inbox/Collections-ct.1058.mcz ==================== Summary ==================== Name: Collections-ct.1058 Author: ct Time: 2 December 2023, 10:54:28.858859 pm UUID: 544526d5-2714-6d44-8bfa-54429ce1ecf8 Ancestors: Collections-ct.1056 When clicking on a TextInspectIt, do not send #inspect to the result as it might be a proxy. This aligns with the implementation of #inspectIt (note to myself: we should also add an explaining comment there). =============== Diff against Collections-ct.1056 =============== Item was changed: ----- Method: TextInspectIt>>actOnClickFor:in:at:editor: (in category 'mouse events') ----- actOnClickFor: anObject in: aParagraph at: clickPoint editor: editor "Note: evalString gets evaluated IN THE CONTEXT OF anObject -- meaning that self and all instVars are accessible" Project current addDeferredUIMessage: [ + ToolSet inspect: "do not send #inspect because the result might be a proxy" + (self target + ifNil: [(Compiler evaluate: evalString for: anObject)] + ifNotNil: [:object | object]) ]. - self target - ifNil: [(Compiler evaluate: evalString for: anObject) inspect] - ifNotNil: [:object | object inspect] ]. ^ true !

2 1

The Trunk: Morphic-mt.2145.mcz
by commits＠source.squeak.org 04 Dec '23

04 Dec '23

Marcel Taeumel uploaded a new version of Morphic to project The Trunk: http://source.squeak.org/trunk/Morphic-mt.2145.mcz ==================== Summary ==================== Name: Morphic-mt.2145 Author: mt Time: 4 December 2023, 9:10:32.014721 am UUID: e96bb924-174b-e244-9121-10ca92e1bb6b Ancestors: Morphic-ct.2144 In docking-bar menu's window-menu, adds icon for (older) close operations to better disinguish them besides the (newer) collapse/expand operations...assuming that "close" is the more frequently used one because of it being there for longer. Note that having an icon for every menu item works against that pre-attentive effect of finding more quickly what you are looking for. This also works for, e.g., "one item below the one with that icon". Thus, no need for every entry in a menu to have an icon. Thus, only an icon for the close-operation, not collapse/expand, even though we have icons for that as well. Let's hope that this is a useful trade-off. :-) =============== Diff against Morphic-ct.2144 =============== Item was changed: ----- Method: TheWorldMainDockingBar>>windowMenuFor:on: (in category 'submenu - windows') ----- windowMenuFor: window on: menu + + | closeBox closeBoxWidth | + closeBox := SystemWindow closeBoxImage. + closeBoxWidth := closeBox width. + menu addItem: [ :item | item contents: 'Close' translated ; + icon: closeBox; target: window; selector: #closeBoxHit ]; addItem: [ :item | item contents: 'Expand (or contract)' translated ; + icon: (MenuIcons blankIconOfWidth: closeBoxWidth); target: window; selector: #expandBoxHit ]; addItem: [ :item | item contents: 'Collapse (or expand)' translated ; + icon: (MenuIcons blankIconOfWidth: closeBoxWidth); target: window; selector: #collapseOrExpand ]; addLine; addItem: [ :item | item contents: 'Close all like this' translated ; + icon: closeBox; target: self; selector: #closeAllWindowsLike:; arguments: { window } ]; addItem: [ :item | item contents: 'Close all but this' translated ; + icon: (MenuIcons blankIconOfWidth: closeBoxWidth); target: self; selector: #closeAllWindowsBut:; arguments: { window } ]; addLine; addItem: [ :item | item contents: 'Collapse all like this' translated ; + icon: (MenuIcons blankIconOfWidth: closeBoxWidth); target: self; selector: #collapseAllWindowsLike:; arguments: { window } ]; addItem: [ :item | item contents: 'Collapse all but this' translated ; + icon: (MenuIcons blankIconOfWidth: closeBoxWidth); target: self; selector: #collapseAllWindowsBut:; arguments: { window } ]!

1 0

[BUG] ExternalCommandShell interrupts StartUpList if no stdio is present
by christoph.thiede＠student.hpi.uni-potsdam.de 02 Dec '23

02 Dec '23

Hi Dave, hi all, I noticed that when I start my normal Trunk image on Windows, which normally does not attach any stdio streams, ExternalCommandShell interrupts the startup process with an error from ExternalCommandShell>>start. See the attached debug log. While I understand while this error occurs, I think that it should be handled silently. Most users (including myself a few times ago) will just abandon this debugger, preventing further startup tasks from being run correctly and leaving the image in a partially uninitialized state. Plus, the debugger is slightly annoying, as I am aware indeed whether I started Squeak from a command shell or not. I would prefer an error to be raised only if I actually try to use an ExternalCommandShell. It would be great if something could overthink this. :-) Best, Christoph --- Sent from Squeak Inbox Talk [ExternalCommandShell startup.SqueakDebug.log]

3 4

The Trunk: Regex-Core-ct.86.mcz
by commits＠source.squeak.org 01 Dec '23

01 Dec '23

Christoph Thiede uploaded a new version of Regex-Core to project The Trunk: http://source.squeak.org/trunk/Regex-Core-ct.86.mcz ==================== Summary ==================== Name: Regex-Core-ct.86 Author: ct Time: 1 December 2023, 9:03:54.138752 pm UUID: 461b5d18-91be-b642-9a2d-d8e56e224948 Ancestors: Regex-Core-tpr.85 Replaces send to #asUnicode with #codePoint, as the former should only used in text conversion scenarios. Thanks to Marcel for the note. =============== Diff against Regex-Core-tpr.85 =============== Item was changed: ----- Method: RxsPredicate>>beUnicodeCategory: (in category 'initialize-release') ----- beUnicodeCategory: categoryName | unicodeClass | unicodeClass := self class unicodeClass. self predicate: [:char | + (unicodeClass generalCategoryTagOf: char codePoint) beginsWith: categoryName].! - (unicodeClass generalCategoryTagOf: char asUnicode) beginsWith: categoryName].!

1 0