Squeak-dev June 2018

squeak-dev@lists.squeakfoundation.org

45 participants
152 discussions

The Trunk: TemporaryVariableScopeEditor-eem.3.mcz
by commits＠source.squeak.org 06 Jun '18

06 Jun '18

Eliot Miranda uploaded a new version of TemporaryVariableScopeEditor to project The Trunk: http://source.squeak.org/trunk/TemporaryVariableScopeEditor-eem.3.mcz ==================== Summary ==================== Name: TemporaryVariableScopeEditor-eem.3 Author: eem Time: 6 June 2018, 10:49:12.17059 am UUID: 6c4b939d-5460-44c5-a5c4-f188ec87b8ea Ancestors: TemporaryVariableScopeEditor-eem.2 Update the TempScopeEditor to work with arbitrary bytecode sets. =============== Diff against TemporaryVariableScopeEditor-eem.2 =============== Item was added: + ----- Method: TempScopeEditor class>>scanBlock (in category 'scanning') ----- + scanBlock + ^[:m| + m isQuick not + and: [(m abstractBytecodeMessageAt: m initialPC) selector = #pushNewArrayOfSize: + and: [(m abstractBytecodeMessageAt: (m concretePCForAbstractPC: 2)) selector = #popIntoTemporaryVariable: + and: [(self new forMethod: m) misplacedTempsExist]]]]! Item was changed: ----- Method: TempScopeEditor>>edit (in category 'editing') ----- edit + self misplacedTempsExist ifTrue: - self buildTempMap. - tempMap notEmpty ifTrue: [| mr | mr := method methodReference. self copyMethodMovingTemps. Transcript cr; show: mr actualClass name, ' >> ', mr methodSymbol. method methodClass compile: out contents classified: mr category]! Item was changed: ----- Method: TempScopeEditor>>editNoCompile (in category 'editing') ----- editNoCompile + ^self misplacedTempsExist ifTrue: - self buildTempMap. - ^tempMap isEmpty ifFalse: [self copyMethodMovingTemps. out contents]! Item was added: + ----- Method: TempScopeEditor>>misplacedTempsExist (in category 'testing') ----- + misplacedTempsExist + self buildTempMap. + ^tempMap notEmpty!

1 0

what's the plan for Etoys in trunk?
by Chris Muller 06 Jun '18

06 Jun '18

Hi all, Does anyone know the state of Etoys in trunk? I just noticed that it has brought its own "HtmlDocument" into Squeak, along with some other very-likely-to-be-in-external-packages classes like "ChessBoard" and "PDA". That, and looking at the number of methods by each major package prefix in the image, I think it's begging to be unloaded from the release image, but when I tried to do that from the MC browser, I got an emergency evaluator. - Chris 'Morphic -> 11373 EToys -> 10105 Kernel -> 4151 System -> 4143 Collections -> 3776 MorphicExtras -> 3233 Graphics -> 2880 Tools -> 2842 Monticello -> 1934 Connectors -> 1742 Network -> 1621 Tests -> 1621 ST80 -> 1539 Compiler -> 1383 OSProcess -> 1140 ToolBuilder -> 1126 Sound -> 1099 CollectionsTests -> 1030 Chronology -> 874 Multilingual -> 852 Nebraska -> 775 KernelTests -> 681 RFB -> 667 WebClient -> 611 SMBase -> 610 Files -> 541 XML -> 539 Regex -> 523 SUnit -> 503 Compression -> 468 Balloon -> 465 TrueType -> 424 Help -> 412 Installer -> 401 SMLoader -> 371 Protocols -> 361 Traits -> 349 PreferenceBrowser -> 347 GeoNames -> 334 ConnectorsShapes -> 310 HelpSystem -> 280 PlotMorph -> 267 FSM -> 257 Services -> 229 MorphicTests -> 210 51Deprecated -> 203 NetworkTests -> 188 ShoutCore -> 183 Environments -> 181 MonticelloConfigurations -> 180 ToolsTests -> 166 XMLPullParser -> 154 GetText -> 153 GraphicsTests -> 150 SystemChangeNotification -> 134 TraitsTests -> 128 SUnitGUI -> 127 CsvImporter -> 126 Squeak -> 124 SqueakSSL -> 124 PackageInfo -> 122 45Deprecated -> 92 60Deprecated -> 90 MultilingualTests -> 77 ReleaseBuilder -> 65 UpdateStream -> 65 SystemReporter -> 52 VersionNumber -> 50 46Deprecated -> 31 ST80Tests -> 27 CommandLine -> 25 SUnitTools -> 22 ST80Tools -> 16 ShoutTests -> 14 VersionNumberTests -> 13 CGPrereqs -> 8 MorphicExtrasTests -> 3 39Deprecated -> 3 MonticelloForTraits -> 2 BalloonTests -> 1 TrunkScript -> 0 '

7 10

Linux zip bloat in trunk
by K K Subbu 06 Jun '18

06 Jun '18

Hi, The Linux zip (both 32 and 64-bit) in trunk bloated from 35M in 18019 to 68M in 18021+ while other platform zips remained slim. Is this an oversight? Regards .. Subbu

1 0

The Trunk: Tools-tpr.818.mcz
by commits＠source.squeak.org 06 Jun '18

06 Jun '18

tim Rowledge uploaded a new version of Tools to project The Trunk: http://source.squeak.org/trunk/Tools-tpr.818.mcz ==================== Summary ==================== Name: Tools-tpr.818 Author: tpr Time: 5 June 2018, 4:52:21.020441 pm UUID: 62fb39e0-5fe8-4598-abb5-be5a0a754547 Ancestors: Tools-eem.817 Clean up a rather odd looking clause in these methods; it was checking to see if a certain class exists in the image but then a) naming the parameter in the #ifNil:ifNotNil: with the class name b) then referring to the class by name within the block. It seems a bit weird that the compiler would accept that without complaining about mis-use of a global as a block temp and so on, but evidently it worled. The code generates exactly the same bytecodes. I think it looks a lot less confusing with a 'proper' temp variable name though. =============== Diff against Tools-eem.817 =============== Item was changed: ----- Method: Browser>>messageHelpFor: (in category 'message list') ----- messageHelpFor: aSelector "Show the first n lines of the sources code of the message behind aSelector." | source formatted iconHelp | Preferences balloonHelpInMessageLists ifFalse: [^ nil]. source := self selectedClassOrMetaClass sourceCodeAt: aSelector ifAbsent: [^ nil]. source lineCount > 5 ifTrue: [ | sourceLines | sourceLines := (source asString lines copyFrom: 1 to: 5) asOrderedCollection. sourceLines add: ' [...]'. source := sourceLines joinSeparatedBy: Character cr]. formatted := (Smalltalk classNamed: #SHTextStylerST80) ifNil: [ source asText ] + ifNotNil: [ :textStylerClass | + textStylerClass new - ifNotNil: [ :SHTextStylerST80 | - SHTextStylerST80 new classOrMetaClass: self selectedClassOrMetaClass; styledTextFor: source asText ]. iconHelp := (self messageIconHelpFor: aSelector) ifNotEmpty: [:t | t , Character cr, Character cr]. ^ iconHelp asText append: formatted; yourself! Item was removed: - ----- Method: Context>>inspectorClass (in category '*Tools-Inspector') ----- - inspectorClass - "Answer the class of the inspector to be used on the receiver. Called by inspect; - use basicInspect to get a normal (less useful) type of inspector." - - ^ ContextInspector! Item was changed: ----- Method: Debugger>>messageHelpAt: (in category 'context stack (message list)') ----- messageHelpAt: anIndex "Show the first n lines of the sources code of the selected message." | method source formatted lineCount | Preferences balloonHelpInMessageLists ifFalse: [^ nil]. contextStack size < anIndex ifTrue: [^ nil]. method := (contextStack at: anIndex) method. source := method getSource. formatted := (Smalltalk classNamed: #SHTextStylerST80) ifNil: [ source asText ] + ifNotNil: [ :textStylerClass | + textStylerClass new - ifNotNil: [ :SHTextStylerST80 | - SHTextStylerST80 new classOrMetaClass: method methodClass; styledTextFor: source asText ]. lineCount := 0. source doWithIndex: [:char :index | char = Character cr ifTrue: [lineCount := lineCount + 1]. lineCount > 10 ifTrue: [ formatted := formatted copyFrom: 1 to: index-1. formatted append: ' [...]'. ^ formatted]]. ^ formatted! Item was changed: ----- Method: MessageSet>>messageHelpAt: (in category 'message list') ----- messageHelpAt: anIndex "Show the first n lines of the sources code of the selected message." | reference source formatted lineCount | Preferences balloonHelpInMessageLists ifFalse: [^ nil]. self messageList size < anIndex ifTrue: [^ nil]. reference := self messageList at: anIndex. reference isValid ifFalse: [^ nil]. source := reference compiledMethod getSource. formatted := (Smalltalk classNamed: #SHTextStylerST80) ifNil: [ source asText ] + ifNotNil: [ :textStylerClass | + textStylerClass new - ifNotNil: [ :SHTextStylerST80 | - SHTextStylerST80 new classOrMetaClass: reference actualClass; styledTextFor: source asText ]. lineCount := 0. source doWithIndex: [:char :index | char = Character cr ifTrue: [lineCount := lineCount + 1]. lineCount > 10 ifTrue: [ formatted := formatted copyFrom: 1 to: index-1. formatted append: ' [...]'. ^ formatted]]. ^ formatted! Item was changed: ----- Method: TimeProfileBrowser>>messageHelpAt: (in category 'message list') ----- messageHelpAt: anIndex "Show the first n lines of the sources code of the selected message." | reference source formatted lineCount | Preferences balloonHelpInMessageLists ifFalse: [^ nil]. self messageList size < anIndex ifTrue: [^ nil]. reference := (self methodReferences at: anIndex) ifNil: [ ^nil ]. reference isValid ifFalse: [ ^nil ]. source := reference compiledMethod getSource. formatted := (Smalltalk classNamed: #SHTextStylerST80) ifNil: [ source asText ] + ifNotNil: [ :textStylerClass | + textStylerClass new - ifNotNil: [ :SHTextStylerST80 | - SHTextStylerST80 new classOrMetaClass: reference actualClass; styledTextFor: source asText ]. lineCount := 0. source doWithIndex: [:char :index | char = Character cr ifTrue: [lineCount := lineCount + 1]. lineCount > 10 ifTrue: [ formatted := formatted copyFrom: 1 to: index-1. formatted append: ' [...]'. ^ formatted]]. ^ formatted!

1 0

The Trunk: ShoutCore-tpr.61.mcz
by commits＠source.squeak.org 06 Jun '18

06 Jun '18

tim Rowledge uploaded a new version of ShoutCore to project The Trunk: http://source.squeak.org/trunk/ShoutCore-tpr.61.mcz ==================== Summary ==================== Name: ShoutCore-tpr.61 Author: tpr Time: 5 June 2018, 4:18:00.297486 pm UUID: ffdd7656-d3c0-4017-994b-166224604658 Ancestors: ShoutCore-mt.60 Small performance optimisation for Shout when text is being saved; instead of the nasty #isKindOf: test the need to preserve the text attribute by sending a message not only actually does the same thing but appears to be intended for just this. =============== Diff against ShoutCore-mt.60 =============== Item was changed: ----- Method: SHTextStylerST80>>unstyledTextFrom: (in category 'converting') ----- unstyledTextFrom: aText "Re-implemented so that TextActions are not removed from aText" | answer | answer := super unstyledTextFrom: aText. aText runs withStartStopAndValueDo: [:start :stop :attribs | + (attribs anySatisfy: [:each | each shoutShouldPreserve]) - (attribs anySatisfy: [:each | each isKindOf: TextAction]) ifTrue: [ attribs do: [:eachAttrib | answer addAttribute: eachAttrib from: start to: stop]]]. ^answer!

1 0

halos now cause a MouseLeave event to be sent to its target
by Chris Muller 05 Jun '18

05 Jun '18

In 5.1, I could have a halo on my morph and let it retain keyboard focus. Now, in 5.2, invoking halos sends a MouseLeave event, which my app responds to by removing keyboard focus. Halos are *around* the morph, the mouse pointer is still *in* the (bounds of) morph. I don't think their invocation should cause this event to be signaled... what do you think?

1 0

The Trunk: Help-Squeak-Project-cmm.53.mcz
by commits＠source.squeak.org 05 Jun '18

05 Jun '18

Chris Muller uploaded a new version of Help-Squeak-Project to project The Trunk: http://source.squeak.org/trunk/Help-Squeak-Project-cmm.53.mcz ==================== Summary ==================== Name: Help-Squeak-Project-cmm.53 Author: cmm Time: 5 June 2018, 2:42:27.90827 pm UUID: 3d4c880f-ce29-400f-8f70-11830baf165e Ancestors: Help-Squeak-Project-tpr.52 Update SqueakV46.sources to SqueakV50.sources. =============== Diff against Help-Squeak-Project-tpr.52 =============== Item was changed: ----- Method: SqueakProjectHelp class>>workingWithSqueak (in category 'pages') ----- workingWithSqueak "This method was automatically generated. Edit it using:" "SqueakProjectHelp edit: #workingWithSqueak" ^(HelpTopic title: 'Working with Squeak' contents: 'Working with Squeak Take few minutes and get familiar with the image concept of the Squeak/Smalltalk system. This includes starting and quitting the image, learning where all the objects live, and understanding where your source code resides. 1. Starting and Quitting Like most Smalltalks, the machine-executable portion is a relatively small program known as the "virtual machine" (VM). The VM''s job is to provide services from the physical machine to real Smalltalk objects. Services like input and output. The Smalltalk system, including all data and code, is a system of objects, built from the ground up, and interpreted by this virtual computer. This affords Smalltalk system platform portability. Smalltalk cannot run without the VM. The VM can''t do anything useful except process Smalltalk systems. To start the system, drag the ".image" data file to the VM executable "squeak". There are myriad command-line options for starting the system via the command-line (see squeak --help). By default, the system will open on the screen in a single OS window. To quit a Squeak session, choose ''quit'' from the menu bar. If you save, the image file will be overwritten and resume from that place the next time it''s launched. 2. The Image File Squeak is an environment built in its own objects from the ground up, including one or more end-user applications. All of the objects in the system -- Classes, Dictionaries, Windows, Customers and other objects that make up the Squeak environment are stored in a binary ".image" file. This is the "object-data file" loaded by the VM when Squeak is launched. When an image is started, every object resumes exactly from where it was last saved. 3. The Sources File Smalltalk is traditionally includes the source code in the running system. However, keeping multiple copies of the same source code in all images files is wasteful. Therefore, the source code itself is kept in a read-only .sources file and accessed by all images. The image files merely have pointers into this file, which is read on-the-fly to present original source code. + The code of the base system is stored in the file "SqueakV50.sources". This file does not change except between releases of Squeak. Normally this file should be placed in the folder containing the VM executable. - The code of the base system is stored in the file "SqueakV46.sources". This file does not change except between releases of Squeak. Normally this file should be placed in the folder containing the VM executable. 4. The Changes File The purpose of Squeak is to develop new programs and systems. Code changes to the running system are effective immediately. But since multiple images can be running, they cannot all update the .sources file safely. Therefore, each image file is accompanied by a ".changes" file which contains source code changes for that and only that Smalltalk system.. The changes file is important for project work. It keeps a sequential log of development activity for the purpose of recovering work performed since the last image-save. Any of several events could lead to the need to recover work, including a power-outage or making an erroneous change to code required to keep the system running. The changes file does not consume memory space, so Squeak is able to keep a complete history of all program changes. This makes it easy to examine or even reinstate older versions of methods (see ''versions'' option in browser selector pane). This encourages experimentation, since you can easily revert to the original versions of any set of methods. In extreme cases where sources and/or changes files are not available, the system can still run, and will automatically decompile the bytecode methods in image memory, if necessary, into readable and editable versions of the original source code (only comments and temporary variable names are lost). 5. Transferring Code-Snippets Between Images In addition to the ''save'' command that saves the entire state of the system, the code of individual methods, categories or classes may be ''filed out'' and then filed-in to another image. 6. Packages The code of an entire project is encapsulated by a Package. This allows users to share their code with other users. Code of packages are delineated by the categories of their classes, and methods. The Monticello browser is then used to wrap that code into a Package object which can be saved to a Monticello repository at http://ss3.gemtalksystems.com/ss. Some projects end up using the resources provided by several packages, resulting in a hierarchy of packages that make up a system. Installer can be used to install such systems.!! ]style[(19 225 26 969 17 449 19 596 19 1351 44 189 12 539)b,,i,,i,,i,,i,,i,,i,!!' readStream nextChunkText) key: #workingWithSqueak!

3 2

Help system changes
by tim Rowledge 05 Jun '18

05 Jun '18

I've been trying to clean up the help system a bit; starting with tidying the code side of things to make it easier to edit the content etc. Mostly I want to update some stuff- having an 'official' help page for Squeak 5.2 that tells people they need SqueakV64.sources is probably a bit unhelpful, for example. Whilst digging through this I note some areas that might be improved and simplified. This really isn't a good time to be making huge changes but we can probably tidy up some. 1. Is it really best to have the current Help menu with a general HelpBrowser opening entry and then several menu items that open browsers on content that is within the general help? Yes, sort-of shortcuts, no, potential for confusion and drift 2. The general help is opened such that the content is displayed in a plain PluggableTextMorphPlus. Any pages displayed within the general help are plain text. However, the 'TerseGuide' and 'Useful Expressions' specific entries open help browsers that are set up with Shout styling. There's a couple issues with that a) The content looks different in the two versions, which seems like a good way to engender some confusion in new users. b) There really isn't any meaningful benefit I can see to using shout styling in either of these two places (I'm sure there are case where it would be important) c) The choice is made in the the TheWorldMainDockingBar>>#commandKeyHelp and related methods, whereas it surely ought to be at least specific to the Help book, better yet specific to the topic and probably best of all, limited to sections of pages where it is marked up to be styled. For 1. I propose reducing the menu entries to Squeak Help, Release Notes & About Squeak. The only complication I see is adding the swiki based help book into the general help. For 2. I suggest dropping the TheWorldMainDockingBar>>#openHelp:topic:styled: and just opening the general help browser. In the future I'd hope we can provide a markup facility to allow a section of text to be styled with Shout or indeed some other mechanism. Maybe we could extend the cmd-6 menu to add that. Other ideas welcomed. tim -- tim Rowledge; tim(a)rowledge.org; http://www.rowledge.org/tim Strange OpCodes: UDF: Use Disk for Frisbee

3 6

Re: [squeak-dev] Don't rely on weak reference cleanup for lookup of undeclared references
by Max Leske 05 Jun '18

05 Jun '18

Bump. > Hi Eliot, > > > > > On 28 May 2018 at 22:03:24, > squeak-dev-request(a)lists.squeakfoundation.org > (squeak-dev-request(a)lists.squeakfoundation.org) wrote: > > Hi Max, > > >> On May 27, 2018, at 2:44 PM, Max Leske <maxleske(a)gmail.com> wrote: >> >> Hi, >> >> Once again, there's a situation in the Fuel test suite that uncovered >> an issue. Consider the following script: >> >> Object subclass: #Foo >> instanceVariableNames: '' >> classVariableNames: '' >> poolDictionaries: '' >> category: 'Kernel-Classes'. >> >> 10 timesRepeat: [(Smalltalk at: #Foo) rename: #Bar. >> (Smalltalk at: #Bar) rename: #Foo ] >> >> In 5.2-alpha this will open a warning dialog, as at some point one of >> the class names will be undeclared. Class>>rename: only checks for an >> entry in the undeclared dictionary without considering that the weak >> references may not yet have been cleaned up. >> >> I understand that this is not really a realistic real world use-case >> but it's still annoying for tests. My current workaround is a forced >> GC after *every* test in the entire suite (I use >> ClassFactoryForTestCase, so class names may be reused many times >> over). Luckily, the impact on >> runtime isn't too bad. >> >> I'd appreciate it if this could be fixed. I'd be even more happy if I >> could be notified of the fix, so that I can remove the workaround >> again. > > Do you have ideas about where the bug lies? > Sort of. In Pharo it seems that when renaming a class, the original > class name is never added to Undeclared while in Squeak it is. > > It looks to me, that the problem is that Environment>>unbind: is never > sent. In Environment>>renameClass:from:to: #binding:removedFrom: > undeclares the original class name and #binding:addedTo: adds the new > binding to the environment. But the now undeclared original class name > remains undeclared. The following modification gets rid of the > problem: > > <snip> > > aClass updateMethodBindingsTo: newBinding. > > declarations add: newBinding. > > self binding: newBinding addedTo: self. > > self unbind: oldBinding key. "<---------------------- added" > > <snip> > > > > HTH, > > Max > > > > >> >> Cheers, >> Max

2 2

The Trunk: Kernel-ul.1176.mcz
by commits＠source.squeak.org 04 Jun '18

04 Jun '18

Levente Uzonyi uploaded a new version of Kernel to project The Trunk: http://source.squeak.org/trunk/Kernel-ul.1176.mcz ==================== Summary ==================== Name: Kernel-ul.1176 Author: ul Time: 4 June 2018, 7:28:14.840591 pm UUID: 70ad3394-d256-4a05-91aa-b5b2937739e3 Ancestors: Kernel-ul.1175 - iterative version of Context >> #copyTo: instead of a recursive one - optimized version of Integer >> #timesRepeat: =============== Diff against Kernel-ul.1175 =============== Item was changed: ----- Method: Context>>copyTo: (in category 'query') ----- copyTo: aContext + "Copy self and my sender chain down to, but not including, aContext. End of copied chain will have nil sender. Assume that there is no loop in the context chain." - "Copy self and my sender chain down to, but not including, aContext. End of copied chain will have nil sender." + | currentContext senderContext copy | - | copy | self == aContext ifTrue: [ ^nil ]. + currentContext := copy := self copy. + [ + senderContext := (currentContext sender ifNil: [ ^copy ]) copy. + currentContext privSender: senderContext. + currentContext := senderContext ] repeat! - copy := self copy. - sender ifNotNil: [ copy privSender: (sender copyTo: aContext) ]. - ^copy! Item was changed: ----- Method: Integer>>timesRepeat: (in category 'enumerating') ----- timesRepeat: aBlock "Evaluate the argument, aBlock, the number of times represented by the receiver." + | remaining | + remaining := self. + [ (remaining := remaining - 1) >= 0 ] whileTrue: [ + aBlock value ]! - | count | - count := 1. - [count <= self] - whileTrue: - [aBlock value. - count := count + 1]!

1 0

Jump to page: