Squeak-dev March 2020

squeak-dev@lists.squeakfoundation.org

38 participants
319 discussions

The Inbox: Compiler-ct.425.mcz
by commits＠source.squeak.org 27 Mar '20

27 Mar '20

Christoph Thiede uploaded a new version of Compiler to project The Inbox: http://source.squeak.org/inbox/Compiler-ct.425.mcz ==================== Summary ==================== Name: Compiler-ct.425 Author: ct Time: 27 March 2020, 8:10:54.53456 pm UUID: fc2a2b1f-76bc-3f4e-b971-65899bec23fb Ancestors: Compiler-ct.424 Fixes a bug regarding decompilation of literal variables The following did not work before: (Object >> #asOrderedCollection) decompile generate valueWithReceiver: 42 arguments: #(). "Error: internal compiler error; should not happen" Please review. Looking at the result of a manual compilation, I assume that literal variables such as {OrderedCollection} should be decompiled as LiteralVariableNodes. I'm however not 100% sure that this does not break anything other. =============== Diff against Compiler-ct.424 =============== Item was changed: ----- Method: DecompilerConstructor>>codeAnyLitInd: (in category 'constructor') ----- codeAnyLitInd: association + ^LiteralVariableNode new - ^VariableNode new name: association key key: association + index: nil - index: 0 type: LdLitIndType!

1 0

The Trunk: EToys-eem.388.mcz
by commits＠source.squeak.org 27 Mar '20

27 Mar '20

Eliot Miranda uploaded a new version of EToys to project The Trunk: http://source.squeak.org/trunk/EToys-eem.388.mcz ==================== Summary ==================== Name: EToys-eem.388 Author: eem Time: 27 March 2020, 10:34:37.924169 am UUID: 637d866a-2ec0-4cc4-a31d-b3ebefeb78be Ancestors: EToys-eem.387 Eliminate another underscore assignment that causes the decompilation tests to error. =============== Diff against EToys-eem.387 =============== Item was changed: ----- Method: PluggableTabBarMorph>>mouseDown: (in category 'actions') ----- mouseDown: anEvent | xPosition newTab | xPosition := anEvent cursorPoint x. + newTab := + (self tabs detect: [ :anAssociation | | tabBounds | - newTab _ - ((self tabs detect: [ :anAssociation | | tabBounds | tabBounds := anAssociation key bounds. (tabBounds left <= xPosition) and: [ tabBounds right >= xPosition]] + ifNone: [^self]) + key. - ifNone: [nil]) - key). newTab ifNil: [^ self]. + newTab = activeTab ifFalse: [ self activeTab: newTab ]! - newTab = activeTab ifFalse: [ self activeTab: newTab ] - !

1 0

The Trunk: Compiler-ct.424.mcz
by commits＠source.squeak.org 27 Mar '20

27 Mar '20

Eliot Miranda uploaded a new version of Compiler to project The Trunk: http://source.squeak.org/trunk/Compiler-ct.424.mcz ==================== Summary ==================== Name: Compiler-ct.424 Author: ct Time: 26 March 2020, 5:19:54.71456 pm UUID: 10f4405b-896c-3149-b76b-b9cd45ca5734 Ancestors: Compiler-nice.420 Fixes a bug regarding decompilation of FullBlockClosures Sample to reproduce: [^ true] decompile should NOT return {[]}! =============== Diff against Compiler-nice.420 =============== Item was changed: ----- Method: Decompiler>>blockTo: (in category 'control') ----- blockTo: end + "Decompile a range of code as in statementsTo:, but return a block node. NB: end is an exclusive index." - "Decompile a range of code as in statementsTo:, but return a block node." | exprs block oldBase lastStatementOfBlockIsNil | oldBase := blockStackBase. blockStackBase := stack size. exprs := self statementsTo: end. lastStatementOfBlockIsNil := pc < method endPC and: [exprs notEmpty and: [exprs last == (constTable at: 4)]]. lastStatementOfBlockIsNil ifTrue: [exprs := exprs allButLast]. block := constructor codeBlock: exprs returns: lastReturnPc = lastPc. blockStackBase := oldBase. lastReturnPc := -1. "So as not to mislead outer calls" ^block! Item was changed: ----- Method: Decompiler>>doClosureCopy:copiedValues: (in category 'control') ----- doClosureCopy: aCompiledBlock copiedValues: blockCopiedValues "implementation note: must be invoked on a copy because it modifies states" | savedPC blockArgs blockTemps blockTempsOffset block mark | numLocalTemps := aCompiledBlock numTemps - aCompiledBlock numArgs - blockCopiedValues size. blockTempsOffset := aCompiledBlock numArgs + blockCopiedValues size. (blockStartsToTempVars notNil "implies we were intialized with temp names." and: [blockStartsToTempVars includesKey: aCompiledBlock]) ifTrue: [tempVars := blockStartsToTempVars at: aCompiledBlock] ifFalse: [blockArgs := (1 to: aCompiledBlock numArgs) collect: [:i| (constructor codeTemp: i - 1 named: 't', (tempVarCount + i) printString) beBlockArg]. blockTemps := (1 to: numLocalTemps) collect: [:i| constructor codeTemp: i + blockTempsOffset - 1 named: 't', (tempVarCount + i + aCompiledBlock numArgs) printString]. tempVars := blockArgs, blockCopiedValues, blockTemps]. tempVarCount := tempVarCount + aCompiledBlock numArgs + numLocalTemps. lastJumpIfPcStack := OrderedCollection new. caseExits := OrderedCollection new. statements := OrderedCollection new: 20. savedPC := pc. self method: (method := aCompiledBlock) pc: aCompiledBlock initialPC. mark := stack size. + block := self blockTo: aCompiledBlock endPC + 1. - block := self blockTo: aCompiledBlock endPC. mark = stack size ifFalse: [self error: 'block did alter the stack']. ^((constructor codeArguments: (tempVars copyFrom: 1 to: aCompiledBlock numArgs) temps: (tempVars copyFrom: blockTempsOffset + 1 to: blockTempsOffset + numLocalTemps) block: block) pc: aCompiledBlock -> savedPC; "c.f. BytecodeEncoder>>pc" yourself).!

1 0

The Inbox: Compiler-eem.422.mcz
by commits＠source.squeak.org 27 Mar '20

27 Mar '20

A new version of Compiler was added to project The Inbox: http://source.squeak.org/inbox/Compiler-eem.422.mcz ==================== Summary ==================== Name: Compiler-eem.422 Author: eem Time: 27 March 2020, 10:26:38.526975 am UUID: dad8c9c5-1392-4977-8f30-cca2d048a71b Ancestors: Compiler-eem.421 Fix a bad bug with Compiler-eem.421. Symbols should not be made writable! Thanks Levente. =============== Diff against Compiler-eem.421 =============== Item was changed: ----- Method: Compiler>>makeCollectionLiteralWritable: (in category 'private') ----- makeCollectionLiteralWritable: lit + "By default all literals are read-only. Usually this is exactly what we desire. Sometimes + this is inconvenient. This method makes collection literals (other than symbols), and + only collection literals, writable, in response to the writableCollectionLiterals cue." - "By default all literals are read-only. Usually this is exactly what we desire. - Sometimes this is inconvenient. This method makes collection literals, and - only collection literals, writable, in response to the writableCollectionLiterals cue" lit isCollection ifTrue: + [lit isSymbol ifFalse: + [lit beWritableObject. + lit isArray ifTrue: + [lit do: + [:subLiteral| self makeCollectionLiteralWritable: subLiteral]]]]! - [lit beWritableObject. - lit isArray ifTrue: - [lit do: - [:subLiteral| self makeCollectionLiteralWritable: subLiteral]]]!

1 0

problem with MenuMorph like morphs
by Ralph Boland 27 Mar '20

27 Mar '20

I have a Morph subclass, say M, that contains a collection of "ItemMorph", say, as submorphs. When an M is displayed the ItemMorph submorphs are displayed in a column. An ItemMorph can act as a display of data but can also act as a button. This code all works as expected. I sometimes need an M morph to act similar to a MenuMorph; that is the M morph can be made to display on command, say from pressing some button. Then once the user presses on one of the M morph submorphs (ItemMorphs) the pressed upon submorph is to like a button invoking some action. Once the selection is made the entire M morph is to be deleted from the screen, like for a MenuMorph. Except in my case the button action never occurs and the M morph is not deleted. To get to the root of the problem I made copies of Classes MenuMorph and MenuItemMorph and was able to get menus to work using these classes. Once I got these copied morphs to work like regular menus I was able to set breakpoints without getting unwanted call stacks. I then made the changes to M and "ItemMorph" based upon what I learned about MenuMorph and MenuItemMorph. I then made a call stack occur when using the copied Morphs by having the method to invoke when a selection is made not exist. My classes M and ItemMorph have every method that is a mehod of MenuMorph or MenuItemMorph on the call stack. But none of them are called when I try to create a menu using M. It seems that some code that handles focus invokes a new process to handle the click by the mouse on a MenuItemMorph but doesn't do so or does so such that nothing happens when I click on a ItemMorph submorph of an M morph on my screen. Where can I set a breakpoint so that I can investigate why nothing happens when I click on a ItemMorph instance submorph of an M morph on the screen? Notes: 1) When my M morph is first displayed there is at the top a balloon that says: Enter text to narrow selection down to matching items I didn't intend for this balloon to occur but it got picked up somehow. It would be useful to know how to change its text or remove it. I point it out here only because it may help explain what is happening. 2) On mouseUp this balloon disappears. 3) after the second mouseDown the M morph gains a shadow and can be picked up and moved. This is an open source project for teaching math/science to young kids. I have problems with the MIT license though so not sure what license I will use. Any help much appreciated and sorry about the post length. Ralph Boland

3 2

Re: [squeak-dev] [cmfcmf/MagicMouse] [Error] Error: could not resolve BaselineOfMagicMouse. (#2)
by Robert 27 Mar '20

27 Mar '20

Alright issue below Kindly, Robert On Fri, Mar 27, 2020 at 11:19, Christoph Thiede <notifications(a)github.com> wrote: > Looks like an issue for the list (squeak-dev) again, what do you think? > > — > You are receiving this because you were mentioned. > Reply to this email directly, [view it on GitHub](https://github.com/cmfcmf/MagicMouse/issues/2#issuecomment-60505754…, or [unsubscribe](https://github.com/notifications/unsubscribe-auth/AH7NNSSNOON5….

1 0

smalltalkCI Trunk broken: vmParameterAt:put: failed
by Jakob Reschke 27 Mar '20

27 Mar '20

Hello, Christoph's idea of having alerts whenever the Trunk builds in smalltalkCI get broken sounds like a good one. @Fabio is it possible to set up a dummy repository that does not really load anything, but runs the Squeak base test suite? Or load an empty dummy package and run the test suite of a different package? This time something goes wrong while setting a VM parameter. Unfortunately it does not tell you where it really happens: Preparing Squeak64-trunk image for CI... ................................................Error: vmParameterAt:put: failed https://travis-ci.org/hpi-swa/Squot/jobs/659875037 (You only see the error in the raw log towards the bottom: https://api.travis-ci.org/v3/job/659875037/log.txt, but there is no additional info to what I pasted above) Kind regards, Jakob

6 22

[ParrotTalk] new release, better ThunkStack integration
by Robert 26 Mar '20

26 Mar '20

Hey y'all, I have just published a new version of ParrotTalk, running on top of ThunkStack. #CryptoGreen. > TSThunkSession now specifies protocol, cipher and encoder. So subclasses can default. With ParrotTalk loaded, use > > Session newOnHost: <host> port: <port> and it defaults to ::: > > ^ Session newOnHost: <host> port: <port> protocol: 'ParrotTalk v3.7' cipher: 'AESede' encoder: 'bytes' With Registers, Cryptography and ParrotTalk tests, there are 309 tests that pass green. Without SSL loaded. Still need ParrotTalk loaded, however. I am working out a registration from each package rather than hardcoding the ParrotTalk SessionOperations in SessionAgentMap, now in ThunkStack. So as each protocol package loads it registers with ThunkStack. Installer ss project: 'Cryptography'; install: 'ProCrypto-1-1-1'; install: 'ProCryptoTests-1-1-1'; install: 'SSLLoader'. -- Kindly, Robert

1 0

smalltalkCI and Squeak 5.3/trunk
by Fabio Niephaus 26 Mar '20

26 Mar '20

Hi all, In case you're using smalltalkCI [1] for testing: It's now using the latest OpenSmalltalkVMs for Squeak 5.3 and trunk builds. I've also updated the trunk images, so these builds should run a bit faster again (less time spent updating the image). Please report any issues at [2]. Cheers, Fabio [1] https://github.com/hpi-swa/smalltalkCI [2] https://github.com/hpi-swa/smalltalkCI/issues

1 0

The Inbox: Compiler-ct.424.mcz
by commits＠source.squeak.org 26 Mar '20

26 Mar '20

Christoph Thiede uploaded a new version of Compiler to project The Inbox: http://source.squeak.org/inbox/Compiler-ct.424.mcz ==================== Summary ==================== Name: Compiler-ct.424 Author: ct Time: 26 March 2020, 5:19:54.71456 pm UUID: 10f4405b-896c-3149-b76b-b9cd45ca5734 Ancestors: Compiler-nice.420 Fixes a bug regarding decompilation of FullBlockClosures Sample to reproduce: [^ true] decompile should NOT return {[]}! =============== Diff against Compiler-nice.420 =============== Item was changed: ----- Method: Decompiler>>blockTo: (in category 'control') ----- blockTo: end + "Decompile a range of code as in statementsTo:, but return a block node. NB: end is an exclusive index." - "Decompile a range of code as in statementsTo:, but return a block node." | exprs block oldBase lastStatementOfBlockIsNil | oldBase := blockStackBase. blockStackBase := stack size. exprs := self statementsTo: end. lastStatementOfBlockIsNil := pc < method endPC and: [exprs notEmpty and: [exprs last == (constTable at: 4)]]. lastStatementOfBlockIsNil ifTrue: [exprs := exprs allButLast]. block := constructor codeBlock: exprs returns: lastReturnPc = lastPc. blockStackBase := oldBase. lastReturnPc := -1. "So as not to mislead outer calls" ^block! Item was changed: ----- Method: Decompiler>>doClosureCopy:copiedValues: (in category 'control') ----- doClosureCopy: aCompiledBlock copiedValues: blockCopiedValues "implementation note: must be invoked on a copy because it modifies states" | savedPC blockArgs blockTemps blockTempsOffset block mark | numLocalTemps := aCompiledBlock numTemps - aCompiledBlock numArgs - blockCopiedValues size. blockTempsOffset := aCompiledBlock numArgs + blockCopiedValues size. (blockStartsToTempVars notNil "implies we were intialized with temp names." and: [blockStartsToTempVars includesKey: aCompiledBlock]) ifTrue: [tempVars := blockStartsToTempVars at: aCompiledBlock] ifFalse: [blockArgs := (1 to: aCompiledBlock numArgs) collect: [:i| (constructor codeTemp: i - 1 named: 't', (tempVarCount + i) printString) beBlockArg]. blockTemps := (1 to: numLocalTemps) collect: [:i| constructor codeTemp: i + blockTempsOffset - 1 named: 't', (tempVarCount + i + aCompiledBlock numArgs) printString]. tempVars := blockArgs, blockCopiedValues, blockTemps]. tempVarCount := tempVarCount + aCompiledBlock numArgs + numLocalTemps. lastJumpIfPcStack := OrderedCollection new. caseExits := OrderedCollection new. statements := OrderedCollection new: 20. savedPC := pc. self method: (method := aCompiledBlock) pc: aCompiledBlock initialPC. mark := stack size. + block := self blockTo: aCompiledBlock endPC + 1. - block := self blockTo: aCompiledBlock endPC. mark = stack size ifFalse: [self error: 'block did alter the stack']. ^((constructor codeArguments: (tempVars copyFrom: 1 to: aCompiledBlock numArgs) temps: (tempVars copyFrom: blockTempsOffset + 1 to: blockTempsOffset + numLocalTemps) block: block) pc: aCompiledBlock -> savedPC; "c.f. BytecodeEncoder>>pc" yourself).!

2 1

← Newer
1
...
6
7
8
9
10
11
12
...
32
Older →

Jump to page: