Squeak-dev April 2001

squeak-dev@lists.squeakfoundation.org

165 participants
614 discussions

Understanding the file list (was: )
by danielv＠netvision.net.il 30 Apr '01

30 Apr '01

Hi DeepLow. Anyway, to your query, I'd suggest you bring up the halo on an active file list, go for the debug handle, and inspect the morph. Go digging and you will find the classes you're looking for in no time. Projet DeepLow <deeplowsoftware(a)oreka.com> wrote: > Hi everyone, > first of all we would like to thank all those who helped us on our > previous problems (Squeak VW compatibility), and then we would like to > submit another one. > as part of our masters project we are trying to make a windows > explorer in squeak, and we were > wondering if anyone could help us with file handling. Does anyone know > how the file browser works (the pink window), and which methods it > invokes ? > there are many classes containing the word "file", which ones should > we focus on to make get the file tree of a computer, to open a file, > to execute a file from squeak(if it is possible) and to copy/paste and > delete a file. > Thanks in advance for your help. > > > Projet DeepLow > MaÓtrise Informatique > UniversitÈ de Savoie (France) > Explorateur 3D en SmallTalk > Serna Audrey (madame_dodue(a)hotmail.com) > Fleutot Pierre (foreveroux(a)yahoo.fr) > Josserand Samuel (samjosserands(a)yahoo.fr) > Loiseau Mathieu (MateOstl96(a)aol.com)

1 0

Re: Celeste Addresses?
by danielv＠netvision.net.il 30 Apr '01

30 Apr '01

About the state of the Celeste index - A thread a couple of month ago reviewed some of the same issues. JohnMaloney was involved with the original Babar and mentioned reading the index on demand and storing the index as binary as two ways to make it faster. I've recently changed the index to use binary storage. Unfortunately, this required quite a lot of refactoring and Celeste was being edited at the time and I lost the image, etc, so, the code is somewhere between lost and hopelessly out of date and incompatible. However, a few conclusions I did reach - 1. Binary is indeed faster than ascii. To be make this precise - not having to convert the Integers (msgId, message time, length and offset) is faster. This is as John remembered. The parts of the index that are strings (subject, from, to, etc) remain the same speed (at best, see next point). 2. Checking for CrLfs can be very expensive. Particularly when for strings in a binary file, CrLfStream tries to figure out the right mode anew for each string. 3. Binary done right is still no panacea. Significant improvement, but no big breakthrough. A few ideas that came to mind as result - 1. Organize the index according to categories, so we can read only the current category, as per Lex. 2. Write the index in such a way that we can read only as much of a category as is needed (for example, the latest 200 messages). The conceptually simple way is to write the index backwards (latest first), probably the practical way is more along the lines of holding for each category the offsets of it's index entries. On second thought, the index serves two purposes. It keeps the msgID -> message file offsets mapping, which is pretty expensive to recalculate (scanning the message file). It also caches some useful header info for all messages, which could be calculated on demand instead. So another solution would be to eliminate the msgID, and keep the offsets directly in the categories file (so we need to move the logging mechanism to it, for update performance and safety). This means that nothing more than the message and category files would be needed. The header cache function, should we find it necessary, could be done by a separate file keeping TOC entries for the last 1000~ entries we've had to calculate. My guess is for most activations of Celeste, those just 1000 entries would cover everything the user wants to see. Anyway if not then, the data is reparsed from the message text. Whew. I hadn't planned on all that when I started writing. Poke holes, anybody? Daniel "Lex Spoon" <lex(a)cc.gatech.edu> wrote: > Address book handling would be really excellent. There should at least > be the *option* of keeping the address book locally, so it shouldn't > *require* LDAP. But LDAP is cool, too, if someone wants to put it > together. One issue is what to do if there are multiple address > approaches floating around.... > > My favorite feature, by the way, is to pick up addresses and names > automatically as email is viewed, like the "Big Brother Database (BBDB)" > does. > > Along with address book handling, come a lot of, err, opportunities to > improve the composition window. It's kinda lame sitting in a GUI system > and being forced to edit messages as raw text files. > > FWIW, though, even nicer than an address book would be a faster mail > database. Loading the entire index into memory is pretty slow right now > -- it should only load index entries for the category I'm looking at. > If .unclassified. ceases being a pseudo-category, then so be it. > Nowadays, compacting is very fast, and .unclassified. could simply be > built when a compaction is run. > > -Lex > > > "Stephan B. Wessels" <stephan.wessels(a)sdrc.com> wrote: > > I wrote some address book code for Celeste, including a Netscape > > addrebook importer, but got distracted by paying work and had to suspend > > it. If anyone wants to help we could probably get at least that part > > working in a weekend. > > > > - Steve > > > > Mike Rutenberg wrote: > > > > > As far as I know, everyone is doing this manually right now. I > > > certainly am. I want to change this though. > > > > > > LDAP is something I know nothing about, but might be a good option > > > especially if you have an existing (corporate?) address book. > > > > > > Interfaces to an external address book is also a possibility. > > > > > > There are some very interesting options to use the message index > > > information as a fast automatically collected database of "important" > > > email addresses. This is done previously by JWZ (?) Big Brother > > > Database for emacs mail reading. I do this manually by using the > > > "Participants Filter" to find the email address of my intended > > > recipient. I tried some experiments with this last week but have not > > > finished it. > > > > > > Mike

1 0

Re: Smalltalk & Squeak featured on Slashdot
by Karl Ramberg 30 Apr '01

30 Apr '01

Les Tyrrell wrote: > > ----- Original Message ----- > From: Karl Ramberg <karl.ramberg(a)chello.se> > To: <squeak(a)cs.uiuc.edu> > Sent: Monday, April 30, 2001 1:26 PM > Subject: Re: Smalltalk & Squeak featured on Slashdot > > > > > > > Les Tyrrell wrote: > > > > > > Glyph Lefkowitz wrote: > > > > > > > > On Monday 30 April 2001 03:55, Stephane Ducasse wrote: > > > > > And this would be great to have classes that have less than 200 methods ;). > > > > > In the same idea, I strongly think that string does not have to know > > > > > asURL > > > > > as.... > > > > > asblabla > > > > > > > > > > Because then everything is linked together. > > > > > > > > I doubt that this is feasible w/ squeak, but in a perfect world, asURL > > > > asBlaBla and friends *would* be part of String, but they would also have a > > > > module they were associated with; so you could browse methods by class or by > > > > module. In theory, module dependencies could be automatically computed, > > > > then... > > > > > Wouldn't these thing be easier if there was sideways relation, > > inheritance from several classes ? > > > > Karl > > In this case, I would argue against that- the problem, as I see it, is that these methods > fundamentally do not belong where they are being found. This is a symptom of a bad design, not a > tools issue per se. Where I would argue that tools have a role in this is to discover where such > things are going on, and to make this sort of thing more visible to the programmer in a way that > clearly depicts this as NOT being the preferred manner in which to introduce new capabilities into > the image. That is the angle I am taking with Oasis, as described a bit below. > Ok. The funktionality in Oasis seem really cool. What would the image size be if you ran this after major shrink and deleted all the fluff? Karl > - les > > > > These things are basically possible. > > > > > > If you take a chunk of code which represents the most visible bits of functionality of > > > some prospective module, you will discover within it usage of certain selectors in certain > > > contexts. These can be located, and it is possible to identify these with both their > > > usage and their implementations. Do enough of this, and you can probably assign chunks of > > > selectors ( call them interfaces if you wish ) to both the module which uses them ( > > > exclusively, perhaps ) and the classes which have been extended to support such modules. > > > I developed just such an analytical capability in Oasis for just this sort of task- it was > > > meant to assist in the modularization of lots of non-modular code. When I started it, I > > > thought I would be mainly interested in VisualWorks, but since then the same sorts of > > > problems have begun to plague Squeak to an even greater extent. > > > > > > If you do this, then you have knowledge of which selectors are extensions and which are > > > intrinsic properties of the various "base" classes. You then "know" more clearly what > > > belongs in those "base" classes, and what is fluff- too much fluff is not a good sign. > > > > > > > It's really powerful to be able to add methods to Object, sometimes; but it > > > > would be really neat if you could take those methods *OFF* too :) > > > > > > If you know which ones they are, this is trivial. > > > > > > - les > >

1 0

[updates] 36 for 3.1alpha
by Dan Ingalls 30 Apr '01

30 Apr '01

Folks - Herewith a bunch of updates from SqC over the last month. A number of contributions from the list need to be included as well, and we'll try to get to these over the next couple of weeks. My version of 3.1alpha took these updates just fine, but if you have trouble receiving them, it may help to execute the following which Ted Kaehler sent out a while ago... | existing | existing _ ChangeSorter allChangeSetNames. existing _ existing collect: [:cngSet | cngSet copyReplaceAll: '/' with: '_']. existing _ existing collect: [:each | each initialIntegerOrNil]. (existing select: [:each | each notNil]) asSet do: [:each | SystemVersion current registerUpdate: each] Here are the summaries... - Dan --------------------------- 3924UniUpdates-tk -- Ted Kaehler -- 11 April 2001 Make both Disney internal updates and External updates come from the same folder in the same server. When making an update external, we don't have to move the file. It will keep its same number and name. Two indexes, updates.list and <prefix>updates.list will be in that directory. When the user chooses which 'server group' to fetch updates from, any with a * in its name will use the prefixed updates file. When anupdate is broadcast, the user chooses which 'server group' to write to, and any with a * in its name will use the prefixed updates file." 3925MIExtras-tk -- Ted Kaehler -- 11 April 2001 Characters respond correctly to clone (^ self). Move atRandom up to Collection, and define atRandom: in Set also. Streamline how MethodFinder finds constants for expressions of the type (data1 == 43) or (data1 < 52)." 3926doublePossessive-sw -- Scott Wallace -- 11 April 2001 Allows easy construction of otherwise-awkward 'double-possessive' phrases in classic tiles, by adding a menu-item in the viewer for player-valued slots so that you can directly obtain tile phrases of the form 'holder's playerAtCursor's graphic'. Also removes a number of methods and does various cleanup in some of the terrain traversed." 3927userDefChgSetCat-sw -- Scott Wallace -- 11 April 2001 Adds the ability to maintain user-defined change-set-categories. A new item in the change-set-list menu of a change-sorter, 'category functions...', offers the ability to: * Add a new change-set-category * Delete an existing change-set category * Rename an existing change-set category * Add a change set to an existing change-set-category * Delete a change set from an existing change-set-category Three other recently-added change-set-category-related commands have now also been moved into this new categories submenu. Also fixes the find-change-set command in change sorters such that it only offers change sets that reside in the current category; this circumvents the former problem that a change sorter could have a selected change set which was not in its list. Also a couple of further bug fixes and cleanups relating to change-set categories." 3928cruftRemoval-raa -- Bob Arning -- 12 April 2001 removes some unneeded cruft from ButtonProperties" 3929chgSorterSel-sw -- Scott Wallace -- 13 April 2001 Fixes bugs that could leave a change sorter showing classes and methods that were no longer appropriate after certain change-set-category commands." 3930prefJuntaPhaseOne-sw -- Scott Wallace -- 17 April 2001 Phase One of a series of updates that overhaul Preferences. Here we do a few preparations. If manually loading updates one at a time, you would be well advised to load all three of these updates together. This update remembers existing pref values, then restores all pref to their default values so that those default values can be feed into the new preference objects. Also, all existing Preferences panels are removed. Thanks to Bob Arning for the code that hunts down and deletes all preferences panels present in mvc projects. CAUTION: hand-jimmied." 3931prefJuntaPhaseTwo-sw -- Scott Wallace -- 15 April 2001 Retools Preferences so that each individual preference is represented by a Preference object which holds all its state. Numerous old tables are expunged by the postscript to this update. CAUTION: This fileout has been hand-jimmied" PreferencesPanel deleteAllPreferencesPanels 3932prefJuntaPhaseThree-sw -- Scott Wallace -- 15 April 2001 The third and final part of the Preference junta which establshes Preference objects as the holders of all Preference data. CAUTION: extensively hand-jimmied" 3933activeWorldExport-raa -- Bob Arning -- 19 April 2001 ensure right ActiveWorld if dialog needed during #copySmartRootsExport:" 3934psPrintFix-raa -- Bob Arning -- 19 April 2001 fix postscript printing of ProjectViewMorph (the name inside was not printing)" 3935resizeFix-raa -- Bob Arning -- 19 April 2001 When resizing SystemWindows with fast drag ON, filckering was observed after the mouse was released but before the mouse had moved again. This was due to both Sensor- and Event-style calls in the same interaction. Since the Sensor calls were flushing the event queue, the Event calls were getting the wrong answer" 3936projectPrefFix-sw -- Scott Wallace -- 19 April 2001 Important! Fixes a bug that broke things when you entered a pre-existing project after the Preferences junta." 3937RefStream-tk -- Ted Kaehler -- 19 April 2001 Incorporates Yoshiki Ohshima's fix to allow a ByteArray to hold a DataStream that might contain a String. Users should really use a RWBinaryOrTextStream instead for ReferenceStream and SmartRefStream. Puts up an inform window when a ReadWriteStream is used. Fix for LEDMorph>>step" 3938morePSFont-raa -- Bob Arning -- 20 April 2001 More Postscript fixups: - scale the substitute for ComicBold down a little more so it fits etoy tiles better - improve printing ImageMorphs when they have transparent bits - fix printing of 8 bit forms" 3939nilSelectorWrkrnd-sw -- Scott Wallace -- 20 April 2001 Workaround regarding ill-formed scripts directories in uniclasses that have nil keys." 3940menuIconInViewer-sw -- Scott Wallace -- 20 April 2001 Use the classic Smalltalk 'menu' icon as the icon inviting a user click to bring up a menu (a) in the viewer header (b) on every content row of a viewer. Also, viewer rows for which a menu is not useful no longer show a menu icon. Also, as a debugging aid and to assist in getting preexisting viewers to feel the new look, adds a new item 'relaunch this Viewer' to a viewer's main menu." 3941chgdSelector-sw -- Scott Wallace -- 20 April 2001 Provides a way to find all change sets that contain a change for a given selector, independent of class, whether it be add, delete, or modify; this is particularly handy for tracking down in just which change set a given method got deleted -- which itself is handy because that may be the only place from which you can recapture the lost method via 'versions'. For example, try evaluating (in a 3.1a image): ChangeSorter browseChangeSetsWithSelector: #setSizeAndMakeResizable: At the moment there is no UI to this feature -- you need to invoke it by evaluating expressions such as the example above in a workspace 3942miscChgs-sw -- Scott Wallace -- 20 April 2001 * Honor cmd-f in the message-list pane of Lexicons, to take you to the 'find' feature, as it always has from the category-list pane. * Gets the right selector established for reacting to change in #useGlobalFlaps setting. * Three fixes relating to the auto-select feature in Lexicons. This is still not 100% right, but a noticeable improvement. * Fixes the abilitiy of individual morphs to override what happens when asked to add Genie menu items to the halo menu, by moving the addition of the extra before the Genie items are added to the method that adds the Genie items. * The flap-tab menu is an early beneficiary of the preceding change -- it avoids both the extra line and the items that otherwise would be inserted by Genie." 3943WnldFlickerFix -- Andreas Raab -- 21 April 2001 Fixes a flickering problem in Wonderlands." 3944AtRandomFix -- Andreas Raab -- 11 April 2001 Fixes Set>>atRandom looping infintely if set size was divisable by five, e.g., ((Set new: 3) add: 1; yourself) atRandom locked up more often than not." 3945HideFlap-ar -- Andreas Raab -- 21 April 2001 Make sure viewers go away whenever we start to paint not only when invoked through the painting flap." 3946polygonSubmorphs-raa -- Bob Arning -- 23 April 2001 Fixes: If a PolygonMorph with another morph embedded in it is put into a GeeMail and then taken out, the embedded morph's position is messed up." 3947swikiPassword-raa -- Bob Arning -- 23 April 2001 Add password protection to projects published on a SuperSwiki. There is a new preference #passwordsOnPublish which controls whether you will be prompted for a password when you publish. If this preference is off, your projects will be published without a password, allowing others to publish updated versions. If you attempt to publish a new version of a protected project, it will fail. If this preference is on, you will be prompted for a password with which to protect the project. If the project already exists and the passwords do not match, the publish will fail." Preferences addPreference: #passwordsOnPublish categories: #(publishing) default: false balloonHelp: 'Turn this on if you want to be prompted for a password when publishing a project.'. 3948ADPCMTypos-ar -- Andreas Raab -- 23 April 2001 Fixes typos from converting some indexed to named primitives." 3949FlashGrabFix-ar -- Andreas Raab -- 23 April 2001 Fixes a problem when grabbing flash morphs out of their environment." 3950MouseOverTweaks-ar -- Andreas Raab -- 23 April 2001 Make enter/leave notification accurate for morphs being the current mouse focus in which case we have to check explicitly if the hand is really over the receiver." 3951reinstateSizer-raa -- Bob Arning -- 24 April 2001 reinstate the variable spacing control on SyntaxMorphs" 3952DblClickProjects-ar -- Andreas Raab -- 24 April 2001 The CS unifies the mechanisms for starting up with some project into the (now renamed) ProjectLauncher. The ProjectLauncher will load a project when double clicking on it. Works nicely on Windows but needs to be tested on Mac since I'm uncertain if the VM knows where to find the corresponding image file." 3953gifFix-raa -- Bob Arning -- 25 April 2001 Fix some problems reading GIFs: - local color tables are now supported. - size of outCodes increased to handle larger images - handling of transparentIndex fixed" 3954InvalidationFix-ar -- Andreas Raab -- 25 April 2001 Fixes an invalidation problem" 3955RoundedWorldCorners -- Andreas Raab -- 25 April 2001 Fixes rounded corners for worlds." 3956threadSafe -- Michael Rueger, Andreas Raab -- 26 April 2001 Makes the Transcript thread safe. A fix that got lost for a year... ;-) 3957PointlessWarning-ar -- Andreas Raab -- 26 April 2001 Prevent raising a pointless warning in MovieMorph. No VM but MacOS pre X implements the concept of 'extraVMMemory' so it is quite common and perfectly okay to have extraVMMemory return zero." 3958TickDelay -- Andreas Raab -- 27 April 2001 Make the clock menu popup later to not confuse people." 3960FasterScrolling-di -- Dan Ingalls -- 22 April 2001 Scrolling was not achieving the specified times due to a fixed step rate, but variable scroll rate. These changes use a variable step rate to simplify the code and improve consistency in this regard. The default parameters have also been tweaked for slightly faster scrolling."

1 0

Re: SWT Related point
by Bob Arning 30 Apr '01

30 Apr '01

On Mon, 30 Apr 2001 10:54:38 +0200 Stephane Ducasse <ducasse(a)iam.unibe.ch> wrote: >mechanism that would automatically list >all the new/removed methods in comparison to SqC images. This >would be really interesting/necessary Stephane, Here is something along those lines that I have use in the past. --in the first image-- 1. File in the change set. 2. Evaluate (Smalltalk writeChangeSummary) and wait for it to complete --in the second image-- 3. File in the change set 4. Make sure that the file 'ST80.summary2' from the first image is in the default directlry for this image. 5. Evaluate (Smalltalk compareChangeSummary) When this completes, a workspace will open listing the classes and methods that are present in one image, but not the other or are different between the two images. Note: this is written to compare images of the same major version, i.e. it assumes that methods in the .sources file are the same and doesn't bother checking them. If you are comparing from 2.9 to 3.0, e.g., change the "cm fileIndex > 1" to "cm fileIndex > 0" in two places. Cheers, Bob 'From Squeak3.1alpha [latest update: #''Squeak3.1alpha'' of 5 February 2001 update 3953] on 30 April 2001 at 2:36:22 pm'! !ClassDescription methodsFor: 'fileIn/Out' stamp: 'RAA 4/30/2001 14:31'! compareSummaryOfChangesIn: aDictionary to: aStream | d cm crc otherMethods first printer | first _ true. printer _ [ :msg | first ifTrue: [ aStream cr; nextPutAll: '======================= D======'; cr; nextPutAll: self printString; cr; nextPutAll: '======================= D======'; cr. ]. first _ false. aStream nextPutAll: msg; cr. ]. otherMethods _ aDictionary at: self printString ifAbsent: [ printer value: self printString,' not in other images'. ^self ]. d _ self methodDict. d keys asSortedCollection do: [:sel | cm _ d at: sel. cm fileIndex > 1 ifTrue: [ (otherMethods includesKey: sel asString) ifTrue: [ crc _ (cm getSourceFor: sel in: self) string crc16 printString. crc = (otherMethods at: sel asString) ifFalse: [ printer value: sel,' different in two images'. ]. otherMethods removeKey: sel asString. ] ifFalse: [ printer value: sel,' not in other image'. ]. ]. ]. otherMethods keys asSortedCollection do: [ :k | printer value: k,' in other image, but not in changes'. ]. aDictionary removeKey: self printString. ! ! !ClassDescription methodsFor: 'fileIn/Out' stamp: 'RAA 12/8/2000 19:15'! writeSummaryOfChangesTo: newFile | src d cm | newFile nextPut: $1; nextPutAll: self printString; cr. d _ self methodDict. d keys asSortedCollection do: [:sel | cm _ d at: sel. cm fileIndex > 1 ifTrue: [ src _ cm getSourceFor: sel in: self. newFile nextPut: $2; nextPutAll: sel asString; tab; nextPutAll: src string crc16 printString; cr. ]. ]. ! ! !SystemDictionary methodsFor: 'housekeeping' stamp: 'RAA 4/30/2001 14:31'! compareChangeSummary "Smalltalk compareChangeSummary" | f classCount otherClasses className methods methodName crc endOfClass answer | answer _ WriteStream on: String new. otherClasses _ Dictionary new. f _ FileStream fileNamed: 'ST80.summary2'. [f atEnd] whileFalse: [ f next = $1 ifFalse: [self halt]. className _ f upTo: Character cr. otherClasses at: className put: (methods _ Dictionary new). endOfClass _ false. [f atEnd | endOfClass] whileFalse: [ f next = $2 ifTrue: [ methodName _ f upTo: Character tab. crc _ f upTo: Character cr. methods at: methodName put: crc. ] ifFalse: [ f skip: -1. endOfClass _ true. ]. ]. ]. f close. 'Comparing Changes File...' displayProgressAt: Sensor cursorPoint from: 0 to: Smalltalk classNames size during: [:bar | classCount _ 0. Smalltalk allClassesDo: [ :class | bar value: (classCount _ classCount + 1). class compareSummaryOfChangesIn: otherClasses to: answer. class class compareSummaryOfChangesIn: otherClasses to: answer. ] ]. otherClasses keys asSortedCollection do: [ :k | answer cr; nextPutAll: '======================= D======'; cr; nextPutAll: k,' is not in this image'; cr; nextPutAll: '======================= D======'; cr. ]. StringHolder new contents: answer contents; openLabel: 'differences'. ! ! !SystemDictionary methodsFor: 'housekeeping' stamp: 'RAA 4/30/2001 14:28'! writeChangeSummary "Smalltalk writeChangeSummary" | f classCount | FileDirectory default deleteFileNamed: 'ST80.summary2' ifAbsent: []. f _ FileStream fileNamed: 'ST80.summary2'. 'Summarizing Changes File...' displayProgressAt: Sensor cursorPoint from: 0 to: Smalltalk classNames size during: [:bar | classCount _ 0. Smalltalk allClassesDo: [ :class | bar value: (classCount _ classCount + 1). class writeSummaryOfChangesTo: f. class class writeSummaryOfChangesTo: f ]. ]. f close. ! !

1 0

Re: Smalltalk & Squeak featured on Slashdot
by Karl Ramberg 30 Apr '01

30 Apr '01

Les Tyrrell wrote: > > Glyph Lefkowitz wrote: > > > > On Monday 30 April 2001 03:55, Stephane Ducasse wrote: > > > And this would be great to have classes that have less than 200 methods ;). > > > In the same idea, I strongly think that string does not have to know > > > asURL > > > as.... > > > asblabla > > > > > > Because then everything is linked together. > > > > I doubt that this is feasible w/ squeak, but in a perfect world, asURL > > asBlaBla and friends *would* be part of String, but they would also have a > > module they were associated with; so you could browse methods by class or by > > module. In theory, module dependencies could be automatically computed, > > then... > Wouldn't these thing be easier if there was sideways relation, inheritance from several classes ? Karl > These things are basically possible. > > If you take a chunk of code which represents the most visible bits of functionality of > some prospective module, you will discover within it usage of certain selectors in certain > contexts. These can be located, and it is possible to identify these with both their > usage and their implementations. Do enough of this, and you can probably assign chunks of > selectors ( call them interfaces if you wish ) to both the module which uses them ( > exclusively, perhaps ) and the classes which have been extended to support such modules. > I developed just such an analytical capability in Oasis for just this sort of task- it was > meant to assist in the modularization of lots of non-modular code. When I started it, I > thought I would be mainly interested in VisualWorks, but since then the same sorts of > problems have begun to plague Squeak to an even greater extent. > > If you do this, then you have knowledge of which selectors are extensions and which are > intrinsic properties of the various "base" classes. You then "know" more clearly what > belongs in those "base" classes, and what is fluff- too much fluff is not a good sign. > > > It's really powerful to be able to add methods to Object, sometimes; but it > > would be really neat if you could take those methods *OFF* too :) > > If you know which ones they are, this is trivial. > > - les

1 0

looking for answers to exercises in Squeak book Ch 2.
by txporter＠mindspring.com 30 Apr '01

30 Apr '01

As a new Squeaker, I am slowly making my way through Mark's book on Squeak. At the end of chapter 2, there are some exercises, specifically No. 3: Write a piece of workspace code [...] to do the following: (a) Replace all vowels in the string 'Squeak' with dashes. (b) Compute the average of a set of integers in an array. I did this like so, but think it is more like "procedural-language speak' and not Smalltalk. Anyone have any more elegant solutions? I particularly do not like the use of the aCount variable. aString := 'squeak'. aCount := 1. aString do: [:element | element isVowel ifTrue: [aString at: aCount put: $-]. "Could aString be self?" aCount := aCount +1]. anArray := #(12 14 16 18 20). aSum := 0. aCount := 0. anArray do: [:num | aSum := aSum + num. aCount := aCount +1]. anAverage := aSum / aCount. Sorry for such a trivial start, but I really want to grok the OO way to do this. Tom Porter

10 10

Re: VMMaker (was Re: Stable Squeak?)
by Tim Rowledge 30 Apr '01

30 Apr '01

"phiho.hoang" <phiho.hoang(a)home.com> is widely believed to have written: > After some discussion with Andreas (off line) and reading the Plugin Swiki > (pointer from Andreas) > http://minnow.cc.gatech.edu/squeak/1448, I just updated this a tiny bit to correct the description of the unix socket polling hook. > I think I have a proposal for > _another_ simple, uniform and right (hopefully, with help from all > interested in the VM ;-) way to create a _really_ minimal VM and to handle > the plugins. If you want a _really_ minimal vm, there aren't many things left to move out into plugins:- window/os stuff - ie what is typically in sq{plat}Window.c A lot of it could be in a plugin I suspect. the interpreter - since the jitter is a plugin, maybe the plain interpreter ought to be, so you can hook to whichever one is there. the object memory - probably quite a project to separate out from the plain interpreter, but potentially worthwhile since you could then implement a different model if you need, say, transaction rollback semantics or whatever. On my Acorn, the core vm consists of interp.c, window.c, event.c, argumnenhandling.c and externalprims.c and is 128kb with an inlined interp.c which is quite reasonable. It's also 10% faster than aving everything builtin. One remaining little appendix that couldbe removed someday soon is the pair of vm tables that translate the obsolete indexed and named primitives. There might be 10kb to remove I guess. tim -- Tim Rowledge, tim(a)sumeru.stanford.edu, http://sumeru.stanford.edu/tim Useful random insult:- Both oars in the water, but on the same side of the boat.

1 0

Re: Different Forks for Different Folks
by Richard A. O'Keefe 30 Apr '01

30 Apr '01

Glyph Lefkowitz <glyph(a)twistedmatrix.com> wrote: Although I've never worked in the medical industry, I've worked on custom vertical-market software. This stereotypical cry of frustration, often idealized into the anguished cry of the virtuous user against their technocratic opressors, is equally often just a declaration of resistance to change. "This software engineer knew nothing about how we do our jobs!" can really mean "I knew how to do this job with yellow sticky notes, and the interface doesn't remind me of the yellow sticky notes I know!" It can also mean "This software engineer WASN'T ALLOWED to talk to users". For what it's worth, my local GP's computer system flatly refuses to let them spell my name correctly. My name is "O'Keefe". But their system will not allow anything except letters in names. So they try to enter "OKeefe" which at least lets them get the capitalisation right. But the system knows that names have exactly one capital letter, at the front, and won't let them do that either. Whoever designed the software had not taken the trouble to look at a telephone book. Then there was the court case where I was an expert witness. The software was (a) 3 years late (it was supposed to take 6 months), (b) seriously buggy, (c) atrociously slow, and (d) required the sales-clerk to enter the same information on two screens because the designer KNEW that purchase orders and invoices HAD to be different forms, even though the customer had stated over and over and over again (with examples aplenty) that their parent company required them to use a single form. Then there's the electricity company (Contact Energy) that sent me, as the bill for December 1999, a repeat of the May 1999 bill, whose computer *would not let the customer service operator, or her supervisor, or his supervisor, correct the mistake. Using technology to improve one's job performance or enjoyment does not mean familiarity, it doesn't mean that the software engineer ought to know exactly how the job used to be done; in fact, if the net result of buying an expensive piece of software is that the job gets done in exactly the same way it was done before, but with a $1000 computer and $500 worth of software, why would ANYONE buy business software? Basic usability fact 1: the people who make the purchase decisions are not the people who have to use the software. Basic usability fact 2: the offical work processing rules the analyst is told about are a fiction; the way things _really_ work is different. Basic usability fact 3: mistakes happen; it is important to have reasonable facilities for correcting them. A classic book on data quality estimates the proportion of bad fields in real working commercial data bases as about 20%. The figure for the NZ Police computer system is apparently comparable. Of course you can say "operator error", but _why_ are operators making errors at such rates and _why_ is it so hard to correct them? Of course, software in those situations is a tool, and using a tool requires learning. In my experience, a surprising number of people are not interested in or actively resistant to learning. Why _should_ someone be interested in a tool which was imposed on them? If the fellow who designed that "piece of crap" didn't know anything about how they do their jobs, why did they buy the software? See basic usability facts 1 and 2. I imagine it's not free. If someone ELSE bought the software (not the RN you spoke to), perhaps there are different ideas about how an RN does, or should do, their job. And it was the software vendor's job to find out which one best fits reality. I was reading a book about quality recently (I'm really sorry I can't remember whose, but perhaps someone will remember the anecdote). The author claimed that this really happened: Manager comes to tech support and asks for a laptop computer to take to a weekend conference. Technician explains that they're all out except for one, which needs repairs. Manager insists on having it anyway, then complains that it's too heavy. Light goes on in technician's head. In manager's presence and with manager's informed consent, technician removes battery pack. Manager goes away satisfied. Sometimes a computer is just an executive status symbol.

2 1

Re: Bug in package browser (fwd)
by Scott A Crosby 30 Apr '01

30 Apr '01

On Sat, 28 Apr 2001, Andreas Kuckartz wrote: > I could not reproduce this. > > Which version of Squeak are you using and what is your latest update? > 3.0 VM UNIX. Most recent 3.1 beta image... (all updates).. Not sure how to get the most recent update number out of squeak. Open up a 'package browser', pick any package group: ('System') Then for the pane listing individual packages within that group, middle-click to get a menu, select 'add item'. It will ask if you want to add a package called 'Catagory-Name'. Type anything or just click 'OK'... CRASH into debugger. Scott > > ----- Original Message ----- > From: "Scott A Crosby" <crosby(a)qwes.math.cmu.edu> > To: <squeak(a)cs.uiuc.edu> > Sent: Thursday, April 26, 2001 9:50 PM > Subject: Bug in package browser (fwd) > > > > ---------- Forwarded message ---------- > > Date: Mon, 23 Apr 2001 14:01:45 -0400 (EDT) > > From: Scott A Crosby <crosby(a)qwes.math.cmu.edu> > > To: squeak(a)cs.uiuc.edu > > Subject: Bug in package browser > > > > Go into the package browser, select a system catagory. (Any one.) > > Then, attempt to create a new package. (by clicking in the package pane). > > > > It crashes into the debugger. > > > > The error appears to be here: > > > > selectedSystemCategoryName > > "Answer the name of the selected system category or nil." > > > > systemCategoryListIndex = 0 > > ifTrue: [^nil]. > > packageListIndex = 0 > > ifTrue: [^ self systemCategoryList at: > systemCategoryListIndex]. > > ^ self package , '-' , (self systemCategoryList at: > systemCategoryListIn > > dex) > > > ^^^^^^^^^^^^^^^^ > > > > > > systemCatagoryListIndex is set at 100-something. > > > > > > -- > > > > Scott > > > > > > -- > > No DVD movie will ever enter the public domain, nor will any CD. The last > CD > > and the last DVD will have moldered away decades before they leave > copyright. > > This is not encouraging the creation of knowledge in the public domain. > > > > > > > -- No DVD movie will ever enter the public domain, nor will any CD. The last CD and the last DVD will have moldered away decades before they leave copyright. This is not encouraging the creation of knowledge in the public domain.

2 2

Jump to page: