Squeak-dev October 2006

squeak-dev@lists.squeakfoundation.org

175 participants
305 discussions

The Weekly Squeak Summary No. 3: Oct 15 - Oct 21, 2006
by Michael Haupt 23 Oct '06

23 Oct '06

Dear Squeakers, welcome to the The Weekly Squeak Summary, a weekly report on what's going on in the world of Squeak based on the Weekly Squeak blog. This issue is assembled by a not-yet jetlagged Michael who is looking forward to meeting some Smalltalkers at OOPSLA in Portland, Oregon. Highlights of the week: Ian Piumarta's Pepsi [1] is an exciting environment for implementing statically-compiled dynamic languages, possibly the foundation for a future implementation of Squeak...? The process for working on the planned release of the 3.10 image is being actively discussed [2]. The discussion started out with Göran Krampe's suggestion to make the development process more community-oriented. Entailing the discussion, the Squeak Foundation Board have decided to establish a release team [4], for which volunteers are sought [5]! Last week's news items: [1] Dynamic messages: a tour of Pepsi http://weeklysqueak.wordpress.com/2006/10/16/dynamic-messages-a-tour-of-pep… [2] Process Woes http://weeklysqueak.wordpress.com/2006/10/16/process-woes/ [3] Squeak: toy or instrument? http://weeklysqueak.wordpress.com/2006/10/18/squeak-toy-or-instrument/ [4] Squeak Foundation Board meeting notes - October 18th, 2006 http://weeklysqueak.wordpress.com/2006/10/19/squeak-foundation-board-meetin… [5] Call for release team volunteers http://weeklysqueak.wordpress.com/2006/10/19/call-for-release-team-voluntee… [6] Recent package releases and announcements http://weeklysqueak.wordpress.com/2006/10/19/recent-package-releases-and-an… [7] An optimization example http://weeklysqueak.wordpress.com/2006/10/20/an-optimization-example/ [8] self vs. super http://weeklysqueak.wordpress.com/2006/10/20/self-vs-super/ You can find this and much more on The Weekly Squeak blog, the newsblog dedicated to the world of Squeak: http://news.squeak.org/ Subscribe to the Weekly Squeak RSS feed: http://news.squeak.org/feed/ Greetings from Portland, Michael

1 0

RE: compiled method source pointer cleanup
by Jerome Peace 22 Oct '06

22 Oct '06

Hi Tim, Instead of looking to change the world to deploy what you have sofar why not put the code on mantis and let others see and comment on it. A change of this magnitude is worth testing with other peoples eyes (and hands and keyboards) before committing to. I'm sure both I and Mr Witzel will show interest and there may be others. -------- a slight change of focus --------------------- More thought on the future of sources: As you know from my previous posts I am dedicated to seeing Squeak Soucres unleashed. I've thought about downloading a CD sized source or changes file and remembered my old experiences with large files. (The conclusion: it is always better to divide and conquor than to put all your eggs into large baskets.) I want to see the extention go in the direction of multilevel version sources with the final level being a small "growth tip" changes file. The upper levels would each have a source of its own. To keep any one file from getting too large each source level may consist of several files so you would have version 3 sources. version 3.1 sources (or version 3.1-1 and version 3.1-2 source files where 3.1-2 is the second part continuation of 3.1-1) and a version 3.1.1 changes file which would be the "growth tip." By making sources multipart no one file would have to grow above a reasonable size. There might even be good reason to make that size ~32M to allow some possible backward compatability. The "growth tip" scheme aids distribution. Once you have the upper level sources distributed once you need not do it again. All that needs to be distributed is the image file and the "growth tip" changes file. The changes file will almost always be quite small. I can imaging people blessing you for that. Yours in service, -- Jerome Peace >compiled method source pointer cleanup >tim Rowledge tim at rowledge.org >Thu Oct 19 05:35:05 UTC 2006 wrote: > >I have some code that adds the source reference to the method >properties. Seems to work ok, which is good so far as it goes. > >Problem now is that I'm out of spare braincells on the next stage. >I'm reasonably sure I could work it out eventually but I have a bunch >of actual work to do so I'm asking for help. > >The current source position/file encoding is a little convoluted - >see StandardSourceFileArray> sourcePointerFromFileIndex: index >andPosition: position - in such a way as to make it a bit tricky to >extend the position range. Now we *could* keep using the same >encoding for larger position ranges but it seems a bit silly to >complicate life. I'd like to use something much more extensible >(remember, the source reference is no longer encoded into a fixed >number of bytes at the end of a compiledmethod) like >(sourcefileindex-1) + (position bitShift: 2) >which would happily keep working into large integers. I think we can >feel sure that large integer will allow pointing into any conceivable >file size? > >What we need is a chunk of code that will >do the equivalent of #condenseSources whilst moving from the old to >the new encoding >install the replacement code to use the new encoding without breaking >during the installation >cleanup > >I suppose the good news about losing the old sources as part of 3.9 >is that there is no need to try to make the transfer also follow the >method history chains. That should simplify things a bit. Obviously I >can supply the changesets for the method changes to the brave volunteer. > >There is also a boatload of code needing attacking to clear out usage >of the trailer bytes, explicit assumptions about use of files and the >indices etc but that can come later. > >So, who's feeling brave and adventurous? > >tim > __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

1 0

Using Squeak on iPaq 4150
by Lord ZealoN 22 Oct '06

22 Oct '06

Hi again. I'm trying to get an usable squeak on my PDA(1). Here are the results at the moment without changes. I got SqueakLight.406 from ftp://squeak.cobss.ch/pub/SqueakLight/ Then I copied to my iPaq on a directory Squeak on the storage card with the squeakvm for WinCe, thanks to Yoshiki Ohshima (2) and SqueakV3.sources from the Squeak3.8 image file. Now, the squeak image runs, and don't very bad, but needs some changes. The first problem is the space. In a pocketpc 30MB (10MB squeaklight, 15MB sources) is a lot of space on storage, and RAM. Then, is needed to change the systems fonts, etc...to take a good look in the little screen. And then, we need to change something to the browsers looks good in the 320x200 screen (on PDA with QVGA screens, 640x480 on VGA screens) Squeak is not running at fullscreen because I'm not capable to make the screenshots then, because the Hardware keys don't work (disabled by the VM?) I will post my advances. Here are some screenshots http://lordzealon.com/squeak/01-launching.bmp http://lordzealon.com/squeak/02-launching2.bmp http://lordzealon.com/squeak/03-memory.bmp http://lordzealon.com/squeak/04-world.bmp http://lordzealon.com/squeak/05-worldmenu.bmp http://lordzealon.com/squeak/06-openingbrowser.bmp http://lordzealon.com/squeak/07-browser.bmp http://lordzealon.com/squeak/08-browserfullscreen.bmp (1) HP iPAQ 4150 - 400 mhz processor - 64MB ram - SD card 512MB - QVGA screen (320x200) - OS: PocketPC 2003 (2) http://www.is.titech.ac.jp/~ohshima/squeak/WinCE/ -- ::Mi blog:: http://blog.lordzealon.com Linux-User: #370919

3 2

Endless string storage
by Damien Cassou 22 Oct '06

22 Oct '06

Hi, I'm trying to get the store string of the text included in a 3.9 workspace (the one which begins with "Welcome to Squeak - an open Smalltalk system"). Workspace allInstances third contents storeString But my image is crashed after a few seconds of 100% CPU usage. -- Damien Cassou

1 0

[ANN] Installer websqueakmap (supports Squeak3.7)
by Keith Hodges 21 Oct '06

21 Oct '06

Installer now supports installing Squeakmap packages via http e.g. squeakmap := Installer websqueakmap. squeakmap package: 'Labby & Walker'; install. ----- Note although it provides the same api as the SMSLoader version, 'websqueakmap', is entirely based upon http: and querying http://map.squeak.com This means it will work in Squeak 3.7 and there is no reason why it would not work in other pre SqueakMap2 images or indeed minimal kernal images that do not have SqueakMap or full Monticello installed. so older squeak versions and squeakmap live again! enjoy best regards Keith Send instant messages to your online friends http://uk.messenger.yahoo.com

2 2

Recompile slows down the compiler
by Klaus D. Witzel 21 Oct '06

21 Oct '06

A strange performance problem happens when methods which refer to "global" variables (classVars, for example) are recompiled *after* the variable in question is initialized to a *large* collection. The sequence of message sends is 1) Parser>>variable 2) Encoder>>encodeVariable:sourceRange:ifUnknown: 3) Encoder>>global:name: and a full trace is below. This example results in a #hash (1 to: 917632 do: ... hashMultiply) of a SparseLargeTable(SequenceableCollection) for every occurence of the class variable in the source text. The reason for this particular slowdown is the use of Association>>#hash which hashes its key xOr its value (and the latter is the *very* *large* collection). Any suggestions? Wouldn't it be sufficient to use the name of the class var, instead of the literal variable (the association). /Klaus P.S. reproducible, for example by filing out a method of Unicode class which accesses the GeneralCategory variable, then opening a code browser on that file (I'm using the tools and settings of the current squeak-dev-41.image). -------------------------- VM: Win32 - a SmalltalkImage Image: Squeak3.9gamma [latest update: #7065] SecurityManager state: Restricted: false FileAccess: true SocketAccess: true Working Dir ...\kWitzel\My Documents\SqueakDev Trusted Dir ...\kWitzel\My Documents\SqueakDev\kWitzel Untrusted Dir ...\My Squeak\kWitzel SmallInteger>>hashMultiply Receiver: 37545126 Arguments and temporary variables: low: nil Receiver's instance variables: 37545126 SparseLargeTable(SequenceableCollection)>>hash Receiver: a SparseLargeTable(1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 29 21 2...etc... Arguments and temporary variables: hash: 37545126 i: 670022 iLimiT: 917632 Receiver's instance variables: base: 1 size: 917632 chunkSize: 1024 defaultValue: 0 Association>>hash Receiver: #GeneralCategory->a SparseLargeTable(1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1...etc... Arguments and temporary variables: Receiver's instance variables: key: #GeneralCategory value: a SparseLargeTable(1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1...etc... Dictionary>>scanFor: Receiver: a Dictionary() Arguments and temporary variables: anObject: #GeneralCategory->a SparseLargeTable(1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 ...etc... element: nil start: nil finish: 22 index: nil indexLimiT: nil Receiver's instance variables: tally: 0 array: #(nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil ni...etc... --- The full stack --- SmallInteger>>hashMultiply SparseLargeTable(SequenceableCollection)>>hash Association>>hash Dictionary>>scanFor: - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Dictionary(Set)>>findElementOrNil: Dictionary>>at:ifAbsent: Encoder>>name:key:class:type:set: Encoder>>global:name: [] in Encoder>>encodeVariable:sourceRange:ifUnknown: {[:assoc | varNode := self global: assoc name: name]} [] in Encoder>>lookupInPools:ifFound: {[:assoc | assocBlock value: assoc. ^ true]} Association(Object)>>ifNotNilDo: [] in Encoder>>lookupInPools:ifFound: {[:sym | (class bindingOf: sym) ifNotNilDo: [:assoc | assocBlock value...]} Symbol class>>hasInterned:ifTrue: Encoder>>lookupInPools:ifFound: [] in Encoder>>encodeVariable:sourceRange:ifUnknown: {[(self lookupInPools: name ifFound: [:assoc | varNode := self global: ...]} Dictionary>>at:ifAbsent: Encoder>>encodeVariable:sourceRange:ifUnknown: Parser>>variable Parser>>primaryExpression Parser>>expression Parser>>primaryExpression Parser>>messagePart:repeat: Parser>>expression Parser>>statements:innerBlock: Parser>>method:context:encoder: [] in Parser>>parse:class:noPattern:context:notifying:ifFail: {[methNode := self method: noPattern context: ctxt encoder: (Enco...]} BlockContext>>on:do: Parser>>parse:class:noPattern:context:notifying:ifFail: Compiler>>format:noPattern:ifFail: Compiler>>format:in:notifying:decorated: Compiler class>>format:in:notifying:decorated: [] in PrettyTextDiffBuilder>>split: {[sourceClass prettyPrinterClass format: trimmed in: sourceClass notify...]} BlockContext>>on:do: PrettyTextDiffBuilder>>split: PrettyTextDiffBuilder(TextDiffBuilder)>>sourceString: PrettyTextDiffBuilder(TextDiffBuilder)>>from:to: PrettyTextDiffBuilder class>>from:to:inClass: TextDiffBuilder class>>buildDisplayPatchFrom:to:inClass:prettyDiffs: [] in FileContentsBrowser>>methodDiffFor:class:selector:meta: {[TextDiffBuilder buildDisplayPatchFrom: source to: aString inClass: th...]} BlockContext>>ensure: CursorWithMask(Cursor)>>showWhile: FileContentsBrowser>>methodDiffFor:class:selector:meta: FileContentsBrowser>>extraInfo FileContentsBrowser>>infoViewContents PluggableTextMorph>>getText PluggableTextMorph>>update: [] in FileContentsBrowser(Object)>>changed: {[:aDependent | aDependent update: aParameter]} DependentsArray>>do: FileContentsBrowser(Object)>>changed: FileContentsBrowser>>updateInfoView FileContentsBrowser>>contents PluggableShoutMorph(PluggableTextMorph)>>getText PluggableShoutMorph(PluggableTextMorph)>>update: [] in FileContentsBrowser(Object)>>changed: {[:aDependent | aDependent update: aParameter]} DependentsArray>>do: FileContentsBrowser(Object)>>changed: FileContentsBrowser(Object)>>contentsChanged FileContentsBrowser(CodeHolder)>>contentsChanged ...etc...

1 0

[ANN] SoapCore 1.0 is now on SqueakMap
by Masashi UMEZAWA 21 Oct '06

21 Oct '06

Hi, SoapCore 1.0 has been released on SqueakMap. http://map.squeak.org/package/dab9b621-00d2-41c3-966c-458bf62b8008/autovers… New features: - Shared values support in SOAP encoding - Custom type support in SOAP encoding - Partial support of document/literal style - Tests are fixed to run in Mac, Unix, Windows - Security test included for SOAP decoding, so it is fairly safe - Runs on Squeak 3.8 & 3.9 Enjoy! -- [:masashi | ^umezawa]

1 0

OOPSLA - what's going on?
by Todd Blanchard 21 Oct '06

21 Oct '06

I'm going to the dynamic languages symposium at oopsla on Monday, will arrive Sunday eve, and stay for the BOF Tuesday eve. I have no idea what else is going on. Anybody got any pointers to Smalltalk meets? -Todd Blanchard

3 2

Re: OOPSLA - what's going on?
by andresv＠itsnpt.com 21 Oct '06

21 Oct '06

Guys, Please check here: http://www.cincomsmalltalk.com/CincomSmalltalkWiki/Smalltalk+World+Tour Thanks, Andres. Original Message: ----------------- From: Avi Bryant avi.bryant(a)gmail.com Date: Fri, 20 Oct 2006 14:37:25 -0700 To: squeak-dev(a)lists.squeakfoundation.org Subject: Re: OOPSLA - what's going on? On Oct 20, 2006, at 2:29 PM, Todd Blanchard wrote: > I'm going to the dynamic languages symposium at oopsla on Monday, > will arrive Sunday eve, and stay for the BOF Tuesday eve. I have > no idea what else is going on. > > Anybody got any pointers to Smalltalk meets? I'm also arriving Sunday evening, and will be at the DLS, but unfortunately have to rush back to Vancouver Monday night and will miss the BOF. So, selfishly, I'd love to arrange a Smalltalker's meeting during the lunch break for the DLS (11:30 - 1:00pm on Monday)... Avi -------------------------------------------------------------------- mail2web - Check your email from the web at http://mail2web.com/ .

1 0

Local squeakmap
by Giuseppe Luigi Punzi 21 Oct '06

21 Oct '06

Hi all Since 3 days ago, is dificult to me to connect to internet, so, I would like to know, if exists an option (or if is a crazy idea) to have all squeakmap on local (how size is it), and when i'm on internet,update my "map" cache with new changes. Then, point all my images to a "squeakmap directory",to install them. Thanks. Cheers. -- Giuseppe Luigi Punzi Linux User: #370919 http://blog.lordzealon.com

2 2

← Newer
1
...
7
8
9
10
11
12
13
...
31
Older →

Jump to page: