Squeak-dev July 2012

squeak-dev@lists.squeakfoundation.org

55 participants
189 discussions

The Trunk: Kernel-eem.708.mcz
by commits＠source.squeak.org 25 Jul '12

25 Jul '12

Eliot Miranda uploaded a new version of Kernel to project The Trunk: http://source.squeak.org/trunk/Kernel-eem.708.mcz ==================== Summary ==================== Name: Kernel-eem.708 Author: eem Time: 24 July 2012, 3:28:27.644 pm UUID: 8d0ba08e-9374-4d98-8f6e-93d5fd9fbea2 Ancestors: Kernel-bf.707 Fix CompiledMethod>equivalentTo:. Old implementation assumed RB parse trees which are comparable. =============== Diff against Kernel-bf.707 =============== Item was changed: ----- Method: CompiledMethod>>equivalentTo: (in category 'comparing') ----- + equivalentTo: aCompiledMethod + | parseTree | + ^self = aCompiledMethod + or: [self class == aCompiledMethod class + and: [self numArgs = aCompiledMethod numArgs + and: [self numLiterals = aCompiledMethod numLiterals + and: [parseTree := self decompile. + (parseTree isKindOf: MethodNode) + ifTrue: [parseTree asString = aCompiledMethod decompile asString "Standard compiler"] + ifFalse: [parseTree = aCompiledMethod decompile "RB parse trees, we hope"]]]]]! - equivalentTo: aCompiledMethod - "does not work yet with non-RB parseTrees" - ^ self = aCompiledMethod - or: [self class == aCompiledMethod class - and: [self numArgs == aCompiledMethod numArgs - and: [self decompile = aCompiledMethod decompile]]].!

1 0

Re: [squeak-dev] The Inbox: Tools-fbs.415.mcz
by Frank Shearar 24 Jul '12

24 Jul '12

On 19 July 2012 14:39, <commits(a)source.squeak.org> wrote: > Frank Shearar uploaded a new version of Tools to project The Inbox: > http://source.squeak.org/inbox/Tools-fbs.415.mcz > > ==================== Summary ==================== > > Name: Tools-fbs.415 > Author: fbs > Time: 19 July 2012, 2:39:10.14 pm > UUID: 5cc4c76f-cd59-410c-b5de-8dd427098e6d > Ancestors: Tools-cmm.414 > > Mantis bug 7697: deleting a message from a MessageSet doesn't automatically select another message, breaking the flow when, say, deleting a large number of messages. > > =============== Diff against Tools-cmm.414 =============== > > Item was changed: > ----- Method: MessageSet>>deleteFromMessageList: (in category 'message functions') ----- > deleteFromMessageList: aMessage > "Delete the given message from the receiver's message list" > + | currIdx | > + currIdx := self messageListIndex. > + messageList := messageList copyWithout: aMessage. > + messageList ifNotEmpty: [self messageListIndex: {currIdx. messageList size.} min]! > - > - messageList := messageList copyWithout: aMessage! *cough* Any takers? I dislike using indices, but they're there and they work. I don't mind addressing the using-indices thing in another ticket: 'make MessageSet stop using indices' or something. frank

2 1

The Trunk: Kernel-bf.707.mcz
by commits＠source.squeak.org 24 Jul '12

24 Jul '12

Bert Freudenberg uploaded a new version of Kernel to project The Trunk: http://source.squeak.org/trunk/Kernel-bf.707.mcz ==================== Summary ==================== Name: Kernel-bf.707 Author: bf Time: 24 July 2012, 11:05:11.73 am UUID: 52d2a763-5135-4fb5-9c1c-cdcdd13013a6 Ancestors: Kernel-eem.706 Fix fileout of changeset missing override methods. =============== Diff against Kernel-eem.706 =============== Item was changed: ----- Method: ClassDescription>>fileOutChangedMessages:on:moveSource:toFile: (in category 'fileIn/Out') ----- fileOutChangedMessages: aSet on: aFileStream moveSource: moveSource toFile: fileIndex "File a description of the messages of this class that have been changed (i.e., are entered into the argument, aSet) onto aFileStream. If moveSource, is true, then set the method source pointer to the new file position. Note when this method is called with moveSource=true, it is condensing the .changes file, and should only write a preamble for every method." | org | (org := self organization) categories do: [:cat | | sels | sels := (org listAtCategoryNamed: cat) select: [:sel | aSet includes: sel]. + (moveSource and: [(cat beginsWith: '*') and: [cat endsWith: '-override']]) + ifTrue: ["when condensing sources/changes, preserve overridden methods" - ((cat beginsWith: '*') and: [cat endsWith: '-override']) - ifTrue: [ sels do: [:sel | self printMethodChunkHistorically: sel on: aFileStream moveSource: moveSource toFile: fileIndex]] ifFalse: [ sels do: [:sel | self printMethodChunk: sel withPreamble: true on: aFileStream moveSource: moveSource toFile: fileIndex]]]!

1 0

some quick and dirty sites
by Chris Cunnington 24 Jul '12

24 Jul '12

OK. I'll email to set a time. I'll make some documentation for people to use. Thanks, Chris

2 1

Daily Commit Log
by commits＠source.squeak.org 24 Jul '12

24 Jul '12

Changes to Trunk (http://source.squeak.org/trunk.html) in the last 24 hours: http://lists.squeakfoundation.org/pipermail/packages/2012-July/005428.html Name: Kernel-eem.706 Ancestors: Kernel-dtl.705 Remove redundant code in CompiledMethod>>copyWithTrailerBytes: that worked around an obsolete bug ============================================= http://lists.squeakfoundation.org/pipermail/packages/2012-July/005429.html Name: KernelTests-eem.228 Ancestors: KernelTests-dtl.227 Test method copying, checking for correct copying of pragmas. =============================================

1 0

The Trunk: KernelTests-eem.228.mcz
by commits＠source.squeak.org 23 Jul '12

23 Jul '12

Eliot Miranda uploaded a new version of KernelTests to project The Trunk: http://source.squeak.org/trunk/KernelTests-eem.228.mcz ==================== Summary ==================== Name: KernelTests-eem.228 Author: eem Time: 23 July 2012, 2:04:25.246 pm UUID: b922a656-6736-4895-8c66-0928c77a5a62 Ancestors: KernelTests-dtl.227 Test method copying, checking for correct copying of pragmas. =============== Diff against KernelTests-dtl.227 =============== Item was changed: ----- Method: CompiledMethodComparisonTest>>testHash (in category 'tests') ----- testHash + <timeout: 300> "takes 90 seconds in trunk 4.3 on a 2.2GHz Intel Core i7 MacBook Pro" | ai | ai := CompiledMethod allInstances. ai do: [:a| ai do: [:b| + a = b ifTrue: [self assert: a hash = b hash]]]. + self assert: (ai collect: [:cm| cm hash]) asSet size * 2 >= ai asSet size! - a = b ifTrue: [self assert: a hash = b hash]]]! Item was added: + ----- Method: CompiledMethodTest>>testCopy (in category 'tests - copying') ----- + testCopy + <pragma: #pragma> + | method copy | + method := thisContext method. + self assert: method pragmas notEmpty. + copy := method copy. + self assert: (method equivalentTo: copy). + self assert: method = copy. + self assert: method ~~ copy. + method pragmas do: + [:p| + self assert: p method == method]. + copy pragmas do: + [:p| + self assert: p method == copy]! Item was added: + ----- Method: CompiledMethodTest>>testCopyWithTrailerBytes (in category 'tests - copying') ----- + testCopyWithTrailerBytes + <pragma: #pragma> + | method copy | + method := thisContext method. + self assert: method pragmas notEmpty. + copy := method copyWithTempNames: #('m' 'c'). + self deny: (method equivalentTo: copy). "equivalentTo: decompiles; since temps differ, equivalentTo: should be false." + self deny: method = copy. "copyWithTempNames: changes the length of a method so these are no longer equal." + self assert: method symbolic = copy symbolic. "but their bytecode should be the same" + self assert: method ~~ copy. + method pragmas do: + [:p| + self assert: p method == method]. + copy pragmas do: + [:p| + self assert: p method == copy]!

1 0

The Trunk: Kernel-eem.706.mcz
by commits＠source.squeak.org 23 Jul '12

23 Jul '12

Eliot Miranda uploaded a new version of Kernel to project The Trunk: http://source.squeak.org/trunk/Kernel-eem.706.mcz ==================== Summary ==================== Name: Kernel-eem.706 Author: eem Time: 23 July 2012, 2:03:27.407 pm UUID: 0e281a27-0e2d-42e5-ab2f-11f4cbabe6ca Ancestors: Kernel-dtl.705 Remove redundant code in CompiledMethod>>copyWithTrailerBytes: that worked around an obsolete bug =============== Diff against Kernel-dtl.705 =============== Item was changed: ----- Method: CompiledMethod>>copyWithTrailerBytes: (in category 'copying') ----- copyWithTrailerBytes: trailer "Testing: (CompiledMethod compiledMethodAt: #copyWithTrailerBytes:) tempNamesPut: 'copy end ' " | copy end start penultimateLiteral | start := self initialPC. end := self endPC. copy := trailer createMethod: end - start + 1 class: self class header: self header. 1 to: self numLiterals do: [:i | copy literalAt: i put: (self literalAt: i)]. (penultimateLiteral := self penultimateLiteral) isMethodProperties ifTrue: [copy penultimateLiteral: (penultimateLiteral copy setMethod: copy; + yourself)]. - yourself). - self pragmas do: - [:p| p setMethod: self]]. start to: end do: [:i | copy at: i put: (self at: i)]. ^copy!

1 0

when to start trimming?
by Chris Muller 23 Jul '12

23 Jul '12

> Having said that, we _should_ start a conversation around what will > become Squeak 4.4. > > I'd like to see a greater emphasis on modularity, myself: if not > having Environments in the base image now, then at least identifying > the worst entanglements and working on cutting those. And tests. > Tests, tests, tests. The tools especially need some kind of framework > within which to test. A mock environment permitting one to test lovely > things like #removeClass would be fantastic. (There are building > blocks all over the place, but nothing assembled in the base image.) Another thing which might be low-hanging fruit for the next version is simply cutting out some of our deprecated stuff. - Removing old methods sending isThisEverCalled. - the 311Deprecated and 39Deprecated packages? - unloadability of some of the packages

3 2

SqueakMap question
by David T. Lewis 23 Jul '12

23 Jul '12

I am trying to add a new SqueakMap entry for VMMaker for a 'head' release in order to simplify loading VMMaker in Squeak 4.3. Tim Rowledge added me as co-maintainer on the SqueakMap account yesterday, so I should have the proper access now. I am trying to create the release from Squeak (a SMReleaseBrowser that I open from the SqueakMap Package Loader). When I try to save the new entry, I get an error that seems to indicate that SqueakMap does not think that my dtl account has access to the VMMaker project. But from the map.squeak.org web site, I am shown as co-maintainer of VMMaker. The failure is in SMReleaseBrowser>>savePackageRelease: with a response from the server that includes the string 'This account has no package with id 2e7f103e-22a6-470d-affe-54b1d04ef34a'. I cannot figure out what is wrong here. Maybe it's a caching problem of some sort, since Tim just added me to the project within the last day? Any ideas? Attached is a screen shot of the entry I am trying to make. Thanks, Dave

2 5

Squeak release process - Thanks Frank!
by Chris Cunnington 21 Jul '12

21 Jul '12

Hi Frank, Thanks for volunteering to be the release leader for Squeak 4.4! It's going to be a great image. If you don't mind, I have a few ideas to make the process easier for you. Take those details I gave you and divide them roughly into two groups: things that need to happen to the alpha; and, things to do with the release candidate. It would be good if somebody fixed "Smalltalk appendChangesTo:", as it's broken. That way the image loses a few M and we'll use SqueakV4.4.changes instead of SqueakV4.1.changes. That would be nice. Do the image building in public. Do an image a week. Six revs, it's Labour Day, and we have a Squeak4.4 release. I ran through the image building process over and over with the members of the SOB watching. How about the whole community? There is no reason this needs to be a secret process. Is there? And it will take a lot of weight from you. Let people test, poke, etc. That combined with producing a weekly image shouldn't be too onerous a process. There is one thing you need to keep for yourself, though. Save the desktop pattern as something you choose. You may balk, but it will be satisfying to see YouTube videos with that pattern. It will compensate for the moments you felt abandoned to a thankless task. One more thing. You can release Squeak4.4 or you can be a nice guy. Don't try to do both at once. At some point you'll need to enact executive privilege and say no. There's always Squeak 4.5. Thanks! Chris

3 3

← Newer
1
...
7
8
9
10
11
12
13
...
19
Older →

Jump to page: