Squeak-dev May 2002

squeak-dev@lists.squeakfoundation.org

160 participants
453 discussions

RE: [GOODIE][Modules][Experimental] AnalysisUtils
by danielv＠netvision.net.il 26 May '02

26 May '02

Hm, I was sure I'd sent that to the list... I'm sending it in a separate file to make the ENH tracker see it. PhiHo Hoang <phiho.hoang(a)rogers.com> wrote: > Hi Daniel, > > I wrote: > > > What is the prerequisite for 'AnalysisUtils' ? > > Actually, I meant where can I get 'MoreAnalysis.1.cs' :-) > > Cheers, > > PhiHo. > > -----Original Message----- > From: squeak-dev-admin(a)lists.squeakfoundation.org > [mailto:squeak-dev-admin@lists.squeakfoundation.org] On Behalf Of PhiHo > Hoang > Sent: Saturday, May 25, 2002 8:54 PM > To: squeak-dev(a)lists.squeakfoundation.org > Subject: RE: [GOODIE][Modules][Experimental] AnalysisUtils > > > Hi Daniel, > > I tried to file in your AnalysisUtils and got: > > 'MessageNotUnderstood: deepGlobalReferencesDo:' > > What is the prerequisite for 'AnalysisUtils' ? > > Please find attached the log file. > > Your contribution is greatly appreciated. > > Cheers, > > PhiHo. > > > -----Original Message----- > From: squeak-dev-admin(a)lists.squeakfoundation.org > [mailto:squeak-dev-admin@lists.squeakfoundation.org] On Behalf Of Daniel > Vainsencher > Sent: Saturday, May 25, 2002 10:03 AM > To: squeak-dev(a)lists.squeakfoundation.org > Subject: [GOODIE][Modules][Experimental] AnalysisUtils > > > An application of graph theory to the analysis of modules (and generally > groups of classes). Try th following examples: > > ModuleAnalyzer reportFor: (Module @ #(Squeak Network Applications > MailReader)) on: Transcript. Transcript endEntry. <do-it> > > ModuleAnalyzer reportFor: (Module @ #(Squeak Morphic Demo)) on: > Transcript. Transcript endEntry. <do-it> > > StrongComponentFinder new findStrongComponentsForNodes: (Module @ > #(Squeak Network Applications Morphic)) deepClasses <explore-it> > > from preamble: > > "Change Set: AnalysisUtils > Date: 25 May 2002 > Author: Daniel Vainsencher > > Prerequisites: > MoreAnalysis.1.cs > > Experimental tools to guide the refactoring system (graph theory is our > friend!!). > > First the fundemental algorithms, then the easy interface. > > StrongComponentsFinder implements a cache for class dependency > information, including the inverted dependencies, and two useful graph > algorithms - > findWeakComponents: collectionOfClasses will return a collection of sets > of classes that are dependent on one another. If a module has more than > one of these, we can separate it into these groups and they will be > non-interdependent modules. It will sometimes make sense to have two > weak components in the same module if the same modules use them both, > but it's a good heuristic. > > This is useful because it will reduce the class count in the minimal > system, remove dependencies on the core, and thus expose the core of the > problem, without requiring any refactoring of code... > > findStrongComponentsForNodes: collectionOfClasses will return a > collection of collections of classes that are cyclicly dependent on one > another. These are groups that can only be unloaded atomically, so if > they're big they probably need refactoring. > > Run on any big chunk of the system, the strong components will show you > what needs to be taken apart most urgently. Just find any one class in > there that you think should be optional, and start severing dependencies > on it. > > ModuleAnalyzer reportFor: aModule on: aStream will spill a nice detailed > analysis of your favorite module in terms of the above algorithms, > showing a lot bout what can and should be done... > > favorite usage: > ModuleAnalyzer reportFor: Module root Squeak Morphic on: Transcript. > Transcript endEntry <do it> > > ClassDependencyViewer is just a beginning of something. Help from > someone handy with simple Morphic lists and drag&drop would be much > appreciated. Mail me.."!

1 0

[GOODIE][Modules][Experimental] AnalysisUtils
by Daniel Vainsencher 26 May '02

26 May '02

An application of graph theory to the analysis of modules (and generally groups of classes). Try th following examples: ModuleAnalyzer reportFor: (Module @ #(Squeak Network Applications MailReader)) on: Transcript. Transcript endEntry. <do-it> ModuleAnalyzer reportFor: (Module @ #(Squeak Morphic Demo)) on: Transcript. Transcript endEntry. <do-it> StrongComponentFinder new findStrongComponentsForNodes: (Module @ #(Squeak Network Applications Morphic)) deepClasses <explore-it> from preamble: "Change Set: AnalysisUtils Date: 25 May 2002 Author: Daniel Vainsencher Prerequisites: MoreAnalysis.1.cs Experimental tools to guide the refactoring system (graph theory is our friend!!). First the fundemental algorithms, then the easy interface. StrongComponentsFinder implements a cache for class dependency information, including the inverted dependencies, and two useful graph algorithms - findWeakComponents: collectionOfClasses will return a collection of sets of classes that are dependent on one another. If a module has more than one of these, we can separate it into these groups and they will be non-interdependent modules. It will sometimes make sense to have two weak components in the same module if the same modules use them both, but it's a good heuristic. This is useful because it will reduce the class count in the minimal system, remove dependencies on the core, and thus expose the core of the problem, without requiring any refactoring of code... findStrongComponentsForNodes: collectionOfClasses will return a collection of collections of classes that are cyclicly dependent on one another. These are groups that can only be unloaded atomically, so if they're big they probably need refactoring. Run on any big chunk of the system, the strong components will show you what needs to be taken apart most urgently. Just find any one class in there that you think should be optional, and start severing dependencies on it. ModuleAnalyzer reportFor: aModule on: aStream will spill a nice detailed analysis of your favorite module in terms of the above algorithms, showing a lot bout what can and should be done... favorite usage: ModuleAnalyzer reportFor: Module root Squeak Morphic on: Transcript. Transcript endEntry <do it> ClassDependencyViewer is just a beginning of something. Help from someone handy with simple Morphic lists and drag&drop would be much appreciated. Mail me.."!

2 2

[Modules][Bug] Pooh handles missing after removing B3D?
by danielv＠netvision.net.il 26 May '02

26 May '02

I can't get halos on stuff, I get a walk for because HaloMorph DNU #addPoohHandle:. This gets me to thinking, when we define class extensions (automatically), there's no check for senders of the method removed, right? I think that's a bug all on it's own... Warning: I'm not absolutely sure that's the precise cause, I noticed this in the most recently stripped image, which is missing quite a few things. OTOH, most removals don't depend on class extensions, and therefore can't remove other classes methods. Daniel Vainsencher

1 0

[ANN] AspectS (0.4.2) & PerspectiveS (0.1.1)
by Robert Hirschfeld 25 May '02

25 May '02

http://www.prakinf.tu-ilmenau.de/~hirsch/Projects/Squeak/AspectS/020525- 042/AspectS.25May2040.cs.gz -- 'From Squeak3.2gamma of 13 January 2002 [latest update: #4743] on 25 May 2002 at 8:40:13 pm'! "Change Set: AspectS Date: 25 May 2002 Author: Robert Hirschfeld (hirschfeld(a)acm.org) Version: 0.4.2 License: Squeak License (http://www.squeak.org/download/license.html) AspectS is an extension of the Squeak metaobject protocol (MOP) to support aspect-oriented programming (AOP). http://www.prakinf.tu-ilmenau.de/~hirsch/Projects/Squeak/AspectS/ WARNING!! Incorrectly using AspectS may result in an inconsistent image that can't be trusted anymore!! Therefore, before experimenting with AspectS, please save and backup your image!! This version of AspectS is based on a port of John Brant's MethodWrappers. http://st-www.cs.uiuc.edu/~brant/Applications/MethodWrappers.html"! -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= http://www.prakinf.tu-ilmenau.de/~hirsch/Projects/Squeak/PerspectiveS/02 0525-011/PerspectiveS.25May2014.cs.gz -- 'From Squeak3.2gamma of 13 January 2002 [latest update: #4743] on 25 May 2002 at 8:14:35 pm'! "Change Set: PerspectiveS Date: 21 May 2002 Author: Robert Hirschfeld (hirschfeld(a)acm.org) Version: 0.1.1 Prerequisite: AspectS 0.4.2 License: Squeak License (http://www.squeak.org/download/license.html) PerspectiveS was inspired by the layers idea of the PIE system by Ira Goldstein and Daniel Bobrow. The main difference between the two systems is that while PIE simultaneously offers alternative designs to developers within their development environment, PerspectiveS allows design alternatives or variations to coexist concurrently at run-time. Such alternatives or variations are active depending on context which can be about anything that can be computed. http://www.prakinf.tu-ilmenau.de/~hirsch/Projects/Squeak/PerspectiveS/ PerspectiveS can be viewed as a management layer on top of AspectS to coordinate aspects and their activation blocks... http://www.prakinf.tu-ilmenau.de/~hirsch/Projects/Squeak/AspectS/"!

1 0

Re: gettimeofday() revisited
by John M McIntosh 25 May '02

25 May '02

>If anyone still things that calling gettimeofday() multiple thousand times per >second is a good idea, here is some top output: Cees have you tried my [ENH] relinquishProcessorForMicroseconds: change? That might improve the CPU usage. Also one of the solutions offered for unix is to use the ITIMER support to have a sig alarm every 1/50 of a sec to update the low resolution clock that is used in the primitive dispatch logic. That isn't turned on my default, for a reason I've not heard yet. On the mac I did try that but had some issues with doing profile sampling so I migrated to a pthread implementation where we do a wait for 16 milliseconds on a semaphore. The accuracy here isn't a big concern, rather we want to know if a larger number of milliseconds have passed, versus a few. Is it possible for you to try either of these solutions? -- -- =========================================================================== John M. McIntosh <johnmci(a)smalltalkconsulting.com> 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ===========================================================================

3 5

Re: RE: morphic slowdown
by Bob Arning 25 May '02

25 May '02

On Sat, 25 May 2002 15:24:58 -0400 "Lex Spoon" <lex(a)cc.gatech.edu> wrote: >Avi Bryant <avi(a)beta4.com> wrote: >> On Sat, 25 May 2002, Bob Arning wrote: >> >> > There may be several factors that add up to the difference you see. >> > One of them is that in 2.8, selecting a class did NOT select a method >> > category and thus showed an empty message list. In 2.9, the '-all-' >> > category gets selected by default and this causes the message list to >> > be populated. Bigger classes take longer to do this. >> >> <sheepish>That would explain that, then</sheepish>. >> Ok, so that benchmark was useless. However, the slowdown is still there >> (honest!); I'll have to find a cleaner way of demonstrating it. > > > >The problem is real regardless of the cause. Maybe we shouldn't select >--all-- by default if it's going to be slowish. > >Incidedentally, my "large lists" patch would probably help here. I >added it to help with very large lists in Celeste and (if I ever got >around to it) in IRC. It should help with the display of any large >lists, however. Avi's worst case under 2.9 was 76 ms, so for a UI issue, I would tend not to worry much. If I were to worry, the first thing I'd do is to make it update the list once rather than twice. Very large lists are another matter, but hopefully these won't happen in the browser. Cheers, Bob

1 0

[FIX] for duplicating 'new class arrived'
by Bergel Alexandre 25 May '02

25 May '02

When loading a segment, check for new class arriving is only performed on the symbol, and not with the module. -- _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._ Bergel Alexandre http://www.iam.unibe.ch/~bergel ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^

1 0

string height?
by Niko Schwarz 25 May '02

25 May '02

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hope this is the right list. All I want to know: For drawing on the HeadingMorph, I'd like to use the drawString: at: method of Canvas. But: the at vastly depends on the string height, isnt it? how can i find out that height for a given string? (i know theres Font height, but how to find the font?) regards, nick - -- The people of Halifax invented the trampoline. During the Victorian period the tripe-dressers of Halifax stretched tripe across a large wooden frame and jumped up and down on it to `tender and dress' it. The tripoline, as they called it, degenerated into becoming the apparatus for a spectator sport. The people of Halifax also invented the harmonium, a device for castrating pigs during Sunday service. -- Mike Harding, "The Armchair Anarchist's Almanac" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.6 (GNU/Linux) Comment: For info see http://www.gnupg.org iD8DBQE877M8c9uyCiO8RNQRAg5cAJ9L5gQ4PkWF8Ra0KHHcwFIQSD31VQCgjCbH aA8FoCzrzMzIbcCv1Yk7h1U= =GHQ8 -----END PGP SIGNATURE-----

2 1

Re: [Modules] Progress report
by danielv＠netvision.net.il 25 May '02

25 May '02

Thanks. Well, you know you're just shooting yourself in the foot - the later you join, the less easy stuff is left... and the EToy is waiting... ;-) Daniel Bergel Alexandre <bergel(a)iam.unibe.ch> wrote: > Hello Daniel, > > I just want to say I appreciate your efforts. I am currently busy for participating actively to the modularization process. Perhaps this summer... > > Regards, > Alexandre > > On Sat, May 25, 2002 at 04:24:45AM +0300, danielv(a)netvision.net.il wrote: > > Someone asked a while ago for a progress report. I just finished running > > everything released that's supposed to unload something, while > > collecting some metrics. The first datum is that in this order, > > everything works off a clean 4857 image. > > > > What was unloaded? > > > > These are small modules, removed based on my work of the last couple of > > weekends. > > #(Squeak Network Applications MailReader) > > #(Squeak Network Applications IRCChat) > > #(Squeak Technology Benchmarks) > > #(Squeak Morphic Demo Games) > > #(Squeak Technology Archives Applications) > > #(Squeak Language Exceptions Tests) > > > > The big ones by Henrik: > > #(Squeak VMConstruction) > > #(Squeak Media Balloon3D) > > > > At this point, these are modules had no incoming dependencies and could > > be removed (I don't know if Henrik worked on this explicitly, I didn't). > > #(Image StandardRelease) > > #(Squeak EToy Imported) > > #(Squeak EToy Outliner) > > #(Squeak EToy UserObjects) > > #(Squeak Development Test TextEditing) > > #(Squeak Development Test SUnit UI) > > #(Squeak Development Test SUnit Tests) > > #(Squeak Development FileList Test) > > #(Squeak Network Applications PluggableWebServer) > > #(Squeak Network UUID) > > #(Squeak Morphic Tests) > > #(Squeak Morphic Widgets) > > #(Squeak Morphic Applications Media) > > #(Squeak Morphic Applications PDA) > > #(Squeak Morphic Library Models) > > #(Squeak Morphic Demo Components Built) > > #(Squeak MVC Symbols) > > #(Squeak Media Movies Player) > > #(Squeak Media Speech PhonemeRecognizer) > > #(Squeak Media Speech Klatt KlattResonatorIndices) > > #(Squeak Media Speech TTS) > > #(Squeak Media Speech) > > #(Squeak Media Balloon TrueType) > > #(Squeak Media Graphics Tools Simplification) > > #(Squeak Media Graphics Tools Triangulation) > > #(Squeak Media Graphics Tools Intersection) > > #(Squeak Media Graphics Tools) > > #(Squeak Media Graphics External) > > #(Squeak Technology FFI Examples X11) > > #(Squeak Technology FFI Examples Win32) > > #(Squeak Technology FFI Examples MacOS) > > #(Squeak Technology FFI Examples) > > #(Squeak Technology FFI Plugin) > > #(Squeak Technology SerialPort) > > #(Squeak Technology Files Tests) > > #(Squeak Language Modules Analysis) > > #(Squeak Language Modules Tests) > > #(Squeak Language Modules Refactorings) > > #(Squeak Language Collections Tests) > > #(Squeak Language Core Tests) > > #(Squeak Language Core Magnitudes Tests) > > > > How far are we? > > Before the unload, we had 1805 classes under Squeak. > > > > After the unload we have 1229 classes under Squeak, of which 804 are > > still strung up in various loops according to a strong connected > > components. That's 804 classes Object is indirectly dependent on - quite > > a lot awaits removal. > > > > Analyzing by major categories, these are the numbers of classes that > > exist: > > Before: > > an OrderedCollection( > > Squeak Language)->208 > > Squeak Technology)->136 > > Squeak Media)->451 > > Squeak MVC)->74 > > Squeak Morphic)->329 > > Squeak Network)->189 > > Squeak Development)->87 > > Squeak VMConstruction)->99 > > Squeak EToy)->232) > > > > And after: > > Squeak Language)->187 > > Squeak Technology)->113 > > Squeak Media)->194 > > Squeak MVC)->72 > > Squeak Morphic)->279 > > Squeak Network)->131 > > Squeak Development)->76 > > Squeak EToy)->175) > > > > This is the script I've used in order to do this (I think I've sent up > > everything except for MoreAnalysis.1.cs which matters only for the SCC > > data, and I haven't yet cleaned up): > > > > ChangeSorter fileIntoNewChangeSet: > > '/home/danielv/offLoadNoteBook/squeak/MovingClasses.cs'. > > ChangeSorter fileIntoNewChangeSet: > > '/home/danielv/offLoadNoteBook/squeak/MoreAnalysis.1.cs'. > > ChangeSorter fileIntoNewChangeSet: > > '/home/danielv/offLoadNoteBook/squeak/CarefulModule.2.cs'. > > > > Preferences disable: #accessOnlineModuleRepositories. > > > > {'plugin-rename.4.cs'. > > 'makeVMConstrReloadable.5.cs'. > > 'dependencyFix.4.cs'. > > 'poolsFix.1.cs'. > > 'fasterModuleSaving-hg.1.cs'. > > 'fixChessPool.3.cs'. > > 'refactoringToolsRefactoring.cs'. > > 'codegeneratorModuleFix.1.cs'} > > do: [:name | |fileName | > > fileName _ '/home/danielv/offLoadNoteBook/squeak/', name. > > ChangeSorter fileIntoNewChangeSet: fileName]. > > > > {'CelesteRemoval.8.cs'. > > 'MailComposition.st'. > > 'RefactoredCeleste.5.cs'. > > 'UnloadingDemos.1.cs'. > > 'DynamicWorldMenu.4.cs'} > > do: [:name | |fileName | > > fileName _ '/home/danielv/offLoadNoteBook/squeak/', name. > > ChangeSorter fileIntoNewChangeSet: fileName]. > > > > ParserRemovedUnusedTemps moveTo: Module @ #(Squeak Language Exceptions > > Core). > > FileList removeSelector: #addFileToZip. > > Benchmarks moveTo: (Module fromPath: #(Squeak Technology Benchmarks) > > forceCreate: true). > > > > | module nextRound | > > #( > > (Squeak Network Applications MailReader) > > (Squeak Network Applications IRCChat) > > (Squeak Technology Benchmarks) > > (Squeak Morphic Demo Games) > > (Squeak Technology Archives Applications) > > (Squeak Language Exceptions Tests) > > ) do: [:path | module _ Module @ path. > > module deepDeclareExternalRefs. > > nextRound _ (module importedModules, {module parentModule}) asSet. > > module repository beStandalone. > > ModuleInstaller upload: module. > > Transcript show: 'About to remove'; show: module; cr. > > ModuleInstaller unload: module. > > Transcript show: 'Now have: '; show: (Module root Squeak submodules > > collect: [:e | e -> e deepClasses size]). > > Transcript show: 'candidates '; show: (nextRound select: [:e | e valid > > and: [(e deepIncomingRefsFromOutside: e) size = 0]])] > > > > | module | > > module := Module @ #(Squeak VMConstruction). > > module defineClassExtensionsOutside: module. > > ModuleInstaller unload: module > > > > | module | > > (module _ Module @ #(Squeak Media Balloon3D)) > > defineClassExtensionsOutside: module. > > ModuleInstaller unload: module > > -- > _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._ > Bergel Alexandre http://www.iam.unibe.ch/~bergel > ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^

1 0

RE: [Win32] VM update (3.2 release candidate)
by John M McIntosh 25 May '02

25 May '02

Sigh. Lets look at the MACROS figure out we've meddled with my copy and dump this again Seems I got a JP_REG define in there which might have lead to confusion as I was integrating Ian's changes. Then again does Ian have a #define JP_REG at the bottom of his sqGnu.h? Note my benchmarks used a hand coded sqGnu.h and tampering with interp.c so I could really what was happening. So without any changes your assemble should look like so. .text .p2align 2,0x90 .L1000: .stabn 68,0,5156,.LM1764-interpret .LM1764: incl %esi movzbl (%esi),%eax movl %eax,56(%esp) .stabn 68,0,5159,.LM1765-interpret .LM1765: addl $4,%edi movl receiver,%eax movl 4(%eax),%eax movl %eax,(%edi) .stabn 68,0,5160,.LM1766-interpret .LM1766: movl 56(%esp),%edx jmp *jumpTable.331(,%edx,4) .p2align 2,0x90 .L1001: .stabn 68,0,5164,.LM1767-interpret .LM1767: incl %esi movzbl (%esi),%ecx movl %ecx,56(%esp) .stabn 68,0,5167,.LM1768-interpret .LM1768: addl $4,%edi movl receiver,%eax movl 8(%eax),%eax movl %eax,(%edi) .stabn 68,0,5168,.LM1769-interpret .LM1769: jmp *jumpTable.331(,%ecx,4) .p2align 2,0x90 .L1002: .stabn 68,0,5172,.LM1770-interpret .LM1770: incl %esi movzbl (%esi),%eax movl %eax,56(%esp) .stabn 68,0,5175,.LM1771-interpret .LM1771: addl $4,%edi movl receiver,%eax movl 12(%eax),%eax movl %eax,(%edi) .stabn 68,0,5176,.LM1772-interpret .LM1772: movl 56(%esp),%edx jmp *jumpTable.331(,%edx,4) .p2align 2,0x90 ok with the jumptableptr void **jumpTablePtr = jumpTable; .text .p2align 2,0x90 .L1000: .stabn 68,0,5156,.LM1764-interpret .LM1764: incl %esi movzbl (%esi),%ebp .stabn 68,0,5159,.LM1765-interpret .LM1765: addl $4,%edi movl receiver,%eax movl 4(%eax),%eax movl %eax,(%edi) .stabn 68,0,5160,.LM1766-interpret .LM1766: jmp *jumpTable.331(,%ebp,4) .p2align 2,0x90 .L1001: .stabn 68,0,5164,.LM1767-interpret .LM1767: incl %esi movzbl (%esi),%ebp .stabn 68,0,5167,.LM1768-interpret .LM1768: addl $4,%edi movl receiver,%eax movl 8(%eax),%eax movl %eax,(%edi) .stabn 68,0,5168,.LM1769-interpret .LM1769: jmp *jumpTable.331(,%ebp,4) .p2align 2,0x90 .L1002: .stabn 68,0,5172,.LM1770-interpret .LM1770: incl %esi movzbl (%esi),%ebp .stabn 68,0,5175,.LM1771-interpret .LM1771: addl $4,%edi movl receiver,%eax movl 12(%eax),%eax movl %eax,(%edi) .stabn 68,0,5176,.LM1772-interpret .LM1772: jmp *jumpTable.331(,%ebp,4) .p2align 2,0x90 .L1003: .stabn 68,0,5180,.LM1773-interpret .LM1773: incl %esi movzbl (%esi),%ebp .stabn 68,0,5183,.LM1774-interpret .LM1774: addl $4,%edi movl receiver,%eax movl 16(%eax),%eax movl %eax,(%edi) .stabn 68,0,5184,.LM1775-interpret .LM1775: jmp *jumpTable.331(,%ebp,4) .p2align 2,0x90 -- -- =========================================================================== John M. McIntosh <johnmci(a)smalltalkconsulting.com> 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ===========================================================================

5 21

← Newer
1
...
4
5
6
7
8
9
10
...
46
Older →

Jump to page: