Squeak-dev December 2000

squeak-dev@lists.squeakfoundation.org

96 participants
272 discussions

Re: [GOODIE] SqueakAmp - an MP3 player for Squeak
by Bert Freudenberg 21 Dec '00

21 Dec '00

On Thu, 21 Dec 2000, Bob Hartwig wrote: > SqueakAmp is an MP3 player for Squeak that has many of the nice > features of WinAmp, and a couple that WinAmp doesn't have. Once you > install SqueakAmp, you'll never again need to leave the comfort of > your Squeak environment to play your music. Hey, this is cool! Thanks! A couple of notes: - some sort of introduction would be helpfull - like adding balloons to buttons. - I clicked on MPEG in the "Available" pane and got a walkback because the "complex contents" is nil - the flopping-out scrollbar moves the whole pane And as for whishes: Streaming support would be nice - I'm tuned into MP3-radio all day (try, for example, http://www.chillambience.org/). But this would have to be supported in the Plugin I guess ... -- Bert

1 0

Whisker Browser v0.91 Released
by Doug Way 21 Dec '00

21 Dec '00

I'm forwarding this first announcement from my Whisker Announcements List to the Squeak list, in case anyone is interested in checking out the new version (0.91), or wants to sign up for the announcements list if they're not already. See below for more information on the Whisker browser, a stacking code browser for Squeak. If you want to sign up for the announcements list, just send me an email. The announcements list is a simple one-way email list on which I announce new releases of Whisker, generally once a month at most. (I won't be posting the announcements to the Squeak list as often.) - Doug Way dway(a)riskmetrics.com, dway(a)mat.net -------- Original Message -------- Subject: Whisker v0.91 Released Date: Sat, 9 Dec 2000 22:06:37 -0500 (EST) From: Doug Way <dway(a)mat.net> To: Whisker Announcements List <dway(a)mat.net> The new version (0.91) of the Whisker Browser for Squeak 2.8 is now available for download! This version contains a number of enhancements and bug fixes since the previous release (0.9), which was the version demoed at the OOPSLA Squeak session in October. To download v0.91 directly, go to: http://www.mindspring.com/~dway/smalltalk/Whisker.0.91-Sq2.8.cs.gz For more information on the Whisker Browser, see: http://www.mindspring.com/~dway/smalltalk/whisker.html Here's the summary list of enhancements and bug fixes added in 0.91, with further details below: Enhancements: 1. Integrated method versions browsing 2. "browse full" menu item in other browsers/inspectors/debuggers will open the method in Whisker (a preference, on by default) 3. Close boxes added to class and method panes 4. Class categories now support an arbitrary number of dashes ("-"), including zero dashes 5. The class hierarchy pane can optionally be switched to show non-Object classes, such as ProtoObject, PseudoContext, etc. 6. Class-side information displayed with a dimmer background color in both class and method panes (a preference, on by default) 7. Most recently selected method pane can be highlighted (a preference, off by default) Bug fixes: - Whisker window now takes Display size into account when first opening, won't open larger than Display - Flop-out scrollbars are no longer sometimes obscured - Class categories with no dashes are no longer broken, due to the categories enhancement above - Hierarchy widget arrows/triangles now blend in with the background color properly - Class pane titles no longer extend outside the title bar - Class panes will always be resizeable on the right side (because of above bug fix) - Selecting a class in the class hierachy no longer expands its children - Fixed instance variable ordering - Cleaned up "no menu yet"s from and added "preferences" to all item menus - Include scrollbar positioning fix for this 2.8 version (already submitted for 2.9alpha base image) - Pretty printing now works Other Miscellaneous Changes: - Left-click on an unselectable list item brings up the item's pop-up menu - Class pane ordering now matches ordering of selections in the class hierarchy - Method categories always alphabetized (won't bother with a preference unless there is demand for one) Further comments on the enhancements: 1. Integrated Versions Browsing - This feature was probably non-essential at this relatively early stage in Whisker's development, but some comments at OOPSLA inspired me to give it a try, and I think it fits in really well with the Whisker layout. If a method has more than one version (i.e. you or someone else has saved it more than once), an expander toggle button triangle will appear next to the method name in the class pane. (The existence of the toggle button can be a handy visual cue to show you which methods you're currently working on, or ones that have changed recently for whatever reason.) Expanding the toggle button will list the method versions below as initials/timestamps. You can select a version and it will appear in a pane in the method stack, with a non-bold selector heading indicating that it is a version, not necessarily the current one (like the existing VersionsBrowser). If you have "diffing" turned on (which it is by default), selecting a single version shows the differences between it and the previous version. The cool part is that if you select any two versions, the diffs will be shown between those versions. For example, you could select the 2nd, 4th and 7th version of method, and it will show the diffs between 2 and 4, the diffs between 4 and 7, and then version 7, all in the method pane stack at the same time. Diffing can be turned off with the "toggle diffing" pop-up menu item. (Do a menu-click on a method version list item in the class pane.) Also, to set it globally, the diffsInChangeList global preference is followed. 2. "Browse full" Opens Methods in Whisker - By default, the "browse full" menu items in various other browsers/inspectors/debuggers will now open a method/class in Whisker instead of a System Browser. (The browseFullOpensWhisker preference lets you turn this off if desired.) Someone else suggested this enhancement, but I was surprised at how useful it was once I implemented it. 3. Close Boxes for Panes - Pretty self-explanatory. They're not as useful with method panes, since they only appear on the top titlebar, unless you turn on the showWhiskerMethodTitles preference which turns on titlebars for all method panes. Still, I prefer just having the single titlebar above the method panes for a particular class... it visually groups them together better (and wastes less space). (Also, I have a plan for having close boxes on every method pane without the need for inserting titlebars, which will be addressed in the future.) 4. Class Categories of Arbitrary Depth - Previously, this was a hack which was hard-coded to expect a single dash in a category name. Now it supports no dashes (one "section"), or even two or more dashes. This should fix a number of bugs people were seeing. 5. Show Non-Object Classes in Hierarchy - This lets you see the non-subclasses of Object such as ProtoObject, PseudoContext, etc. It can be turned on with the pop-up menu in the class hierarchy pane, or by doing a "find class" on something like ProtoObject. It's turned off by default since 98% of the time you don't need to look at those few classes, and they add another indent level to the hierarchy which can make it harder to see things if you have a small display. 6. Class-Side Information in Dimmer Color - This generally looks better if you set the Squeak display depth to 16 or 32 bits, although I tweaked the colors so they at least look reasonable in 8-bit depth mode. This is a preference called classSideShadow which can be turned off. 7. Recently Selected Method Can Be Highlighted - If the preference highlightLatestMethod is turned on, this will display the most recently selected method in a brighter color, which can be handy if you have a large stack of methods. However, it mildly conflicts with #6 above, so I have it turned off by default. That's about it for the new enhancements in 0.91. For the next release, 0.92 (probably due in a couple of months), here is roughly what I have planned: - Fill out basic popup menu functions, such as adding/removing methods, adding/removing categories, etc. - Add support for viewing methods without categories (like "-- all --"). (menu toggle and preference) - Warning when deselecting an edited method - Add titlebars to class categories/class hierarchy panes - Add class menu to class pane titlebar, and category/hierarchy menu to category/hierachy titlebar - the usual bug fixes and small enhancements Also, I will try to release a small changeset sometime sooner to upgrade Whisker 0.91 to work with the latest Squeak 2.9alpha, hopefully within a few weeks. (There have been a bunch of Morphic changes in 2.9alpha. Although I have a partly working upgrade changeset right now... shouldn't be too hard to finish.) I might as well also mention a few things which I hope to address further in the future, sometime after the next (0.92) release. These could potentially constitute a 1.0 release: - "Partial hierarchies" which won't display "uncles" etc. by default - Inspect/explore objects in integrated panes - Minimize/maximize buttons for class and method panes - Close/minimize/maximize buttons on method pane "tabs", not titlebars - Refactoring Browser engine support Apologies for the length of this email, it's probably due to this being the first semi-officially announced release. I plan on posting this message to the Squeak list soon, but in general I won't be posting these release announcements to the Squeak list as often as to this Whisker Announcements List. Comments/feedback are welcome. This announcements list is just a one-way list, though, so if you want to start up some discussion with a larger group, you may want to respond to the Squeak list posting. Anyway, enjoy. - Doug Way dway(a)mat.net

1 0

RE: 2.9 speed
by Peter Crowther 20 Dec '00

20 Dec '00

> From: Henrik Gedenryd [mailto:Henrik.Gedenryd@lucs.lu.se] > I've seen tinyBenchmarks slowdowns on the order of 3-5% from > merely filing in change sets [...] Have you ever seen it speed up? It wouldn't surprise me if the placement of frequently-referenced objects relative to the processor cache made a difference, especially on Intel architectures. tinyBenchmarks will probably fit in cache if all the accesses hit different cache lines, but all you need is for two frequently-accessed addresses to flush each other out of the cache and cause waits while they're re-loaded. I'm not a processor expert, and I know there are several on this list. To the experts: Could this be the difference? And can we watch the cache behaviour on x86 sing the internal monitoring stuff to find out? - Peter

1 0

Re: [FIX] Halo Invalidation Tweak
by Henrik Gedenryd 20 Dec '00

20 Dec '00

Doug Way wrote: > Related to that, at some point it might be interesting to put a fancier > algorithm in DamageRecorder>>recordInvalidRect: which determines whether > damage rectangles should be combined together based on how large they are, how > far apart they are, etc. I once had an idea related to "hinting" morphs about redrawing. Hints about how to invalidate/redraw would be invoked by calling other methods than plain ol' invalidRect:. One such hinted method might do what you want. The important part is that not all morphs would enforce the same invalidation policy. For instance, your strategy seems to fit morphs of a certain nature, like the Halo. Morphs with heavy animation, like bouncing atoms, might like another one. Henrik

1 0

Re: 2.9 speed
by Henrik Gedenryd 20 Dec '00

20 Dec '00

I've seen tinyBenchmarks slowdowns on the order of 3-5% from merely filing in change sets, regardless of what the change sets actually change. This was after loading about 200 updates for 2.9. At about 2990 there was a marked slowdown, but as it turned out, not related to a specific change set (and none of those around 2990+ contained any obvious potential speed-hogs in the code). I explicitly tried different sets with "harmless" changes and this gave a similar slowdown regardless. There was no obvious reason for this (like open change browsers). This may result from loading a large number of updates (e.g. hundreds for 2.9a currently). Perhaps purging a lot of these update change sets (not the actual changes) will give you some speed back? Henrik

1 0

RE: Advice on squeaklets needed
by Bert Freudenberg 20 Dec '00

20 Dec '00

On Tue, 19 Dec 2000 Torsten.Bergmann(a)phaidros.com wrote: > >Another slight correction - ProjectViewMorph does all that for you. So the > >loader script merely becomes this: > > > > ProjectViewMorph > > openFromFile: (FileStream requestURLStream: 'mySqueaklet.extSeg') > > But: > > #openFromFile sends the message #asUnZippedStream to the given string > and this method is not implemented for Strings in the plugin image > I've found at: > (http://wwwisg.cs.uni-magdeburg.de/~bert/squeak/plugin/download.html#win) I just double-checked. The SqueakPlugin.image contained in http://wwwisg.cs.uni-magdeburg.de/~bert/squeak/plugin/npsqueak-noarch.zip does have that method, implemented by ReadWriteStream. -- Bert

1 1

Re: Cincom Open Sources 'Frost'
by merlyn＠stonehenge.com 20 Dec '00

20 Dec '00

>>>>> "Ken" == Ken G Brown <kbrown(a)tnc.com> writes: Ken> I happened across this at <http://www.missionsoft.com/>: Ken> ------ Ken> Mission Software, Inc. has created a Smalltalk Ken> compiler for the Java Virtual Machine. This Ken> compiler allows Smalltalk to run on any JVM. Ken> The compiler currently produces 100% Java Ken> class files fully compatible with the Sun Java Ken> Virtual Machine specification. This allows Ken> Smalltalk and Java code to interact seamlessly Ken> and allows Smalltalk programs to run Ken> anywhere Java runs! The open source Bistro project claims roughly the same thing. -- Randal L. Schwartz - Stonehenge Consulting Services, Inc. - +1 503 777 0095 <merlyn(a)stonehenge.com> <URL:http://www.stonehenge.com/merlyn/> Perl/Unix/security consulting, Technical writing, Comedy, etc. etc. See PerlTraining.Stonehenge.com for onsite and open-enrollment Perl training!

1 0

Less delay - Re: World menu delay -- Re: 2.9 speed
by Bob Arning 20 Dec '00

20 Dec '00

On Mon, 18 Dec 2000 16:01:32 -0800 Duane Maxwell <dmaxwell(a)exobox.com> wrote: >Personally, I find the current behavior annoying, but I'm too polite to say so. > >(Oh, was that out loud?) Duane & all, Loud enough ;-) Here is another candidate for making life a little smoother. Open a new browser (whatever) after filing in to get the full effect. If anyone has strong feeling on this, let's hear about it. Cheers, Bob =====code follows===== 'From Squeak2.9alpha of 17 July 2000 [latest update: #3187] on 18 December 2000 at 7:53:30 pm'! "Change Set: lessDelay Date: 18 December 2000 Author: Bob Arning make pluggable lists a little more responsive: - do not highlight the mouseDown state unless dragging is enabled. - do not gratuitously enable the double-click handler"! !PluggableListMorph methodsFor: 'events' stamp: 'RAA 12/18/2000 19:51'! mouseDown: event onItem: aMorph event yellowButtonPressed ifTrue: [^ self yellowButtonActivity: event shiftPressed]. aMorph ifNotNil: [ self dragEnabled ifTrue: [aMorph highlightForMouseDown] ]! ! !PluggableListMorph methodsFor: 'drag and drop' stamp: 'RAA 12/18/2000 19:51'! installEventHandlerOn: morphList | handler | handler _ EventHandler new. handler on: #mouseDown send: #mouseDown:onItem: to: self; on: #mouseUp send: #mouseUp:onItem: to: self. doubleClickSelector ifNotNil: [ handler on: #doubleClick send: #doubleClick:onItem: to: self. ]. self dragEnabled ifTrue: [handler on: #startDrag send: #startDrag:onItem: to: self]. self dropEnabled ifTrue: [handler on: #mouseEnterDragging send: #mouseEnterDragging:onItem: to: self. handler on: #mouseLeaveDragging send: #mouseLeaveDragging:onItem: to: self]. morphList do: [:m | m eventHandler: handler]. ! !

2 1

World menu delay -- Re: 2.9 speed
by Bob Arning 20 Dec '00

20 Dec '00

Hi all, If the delay in getting a world menu is troubling, try the enclosed change set. The change is that if the were no halos, then the menu pops up right away. If people like this, I will lobby for inclusion. Cheers, Bob On Mon, 18 Dec 2000 16:02:04 -0500 Doug Way <dway(a)riskmetrics.com> wrote: >But I have a feeling that what you're seeing are some 200ms delays which were added to some operations in Morphic. For example, bringing up the World menu now requires that you hold down the button for at least 200ms before it pops up. I think this was done as a simple implementation so that a "quick click" on the desktop would shake off halos. I also notice that this delay is now in PluggableListMorphs too, so that selecting a list item always takes at least 200ms in a Browser, for example. (I'm not sure why this was added here.) > >For what it's worth, I think that these 200ms delays are a VERY BAD idea from a "feel" perspective. It gives the impression that Morphic is slow when it really isn't. (The only case of a delayed pop-up menu in a UI that I can think of is with something like Netscape on the Mac, where if you hold down a mouse on a link, a pop-up menu eventually appears, as opposed to a quick click which simply follows the link. But that delay is much more justifiable than the ones in Squeak, IMHO.) > >I think it should be possible (although slightly more complicated) to implement things so that a 200ms delay in opening the World menu isn't necessary, but halos are still shaken off. I actually implemented a changeset which did most of this a few months ago, but it was made obsolete by the significant changes to Morphic in the last couple months. Maybe I'll take another stab at it, unless someone else wants to try. =====code follows===== 'From Squeak2.9alpha of 17 July 2000 [latest update: #3187] on 18 December 2000 at 4:12:44 pm'! !PasteUpMorph methodsFor: 'event handling' stamp: 'RAA 12/18/2000 16:12'! mouseDown: evt "Handle a mouse down event." | grabbedMorph handHadHalos | grabbedMorph _ self morphToGrab: evt. grabbedMorph ifNotNil:[ grabbedMorph isSticky ifTrue:[^self]. self isPartsBin ifFalse:[^evt hand grabMorph: grabbedMorph]. grabbedMorph _ grabbedMorph partRepresented duplicate. grabbedMorph restoreSuspendedEventHandler. (grabbedMorph fullBounds containsPoint: evt position) ifFalse:[grabbedMorph position: evt position]. "Note: grabbedMorph is ownerless after duplicate so use #grabMorph:from: instead" ^evt hand grabMorph: grabbedMorph from: self]. (super handlesMouseDown: evt) ifTrue:[^super mouseDown: evt]. handHadHalos _ evt hand halo notNil. evt hand halo: nil. "shake off halos" evt hand releaseKeyboardFocus. "shake of keyboard foci" evt shiftPressed ifTrue:[ ^evt hand waitForClicksOrDrag: self event: evt selectors: { #findWindow:. nil. #dragThroughOnDesktop:} threshold: 5]. self isWorldMorph ifTrue: [ handHadHalos ifTrue: [^self addAlarm: #invokeWorldMenu: with: evt after: 200]. ^self invokeWorldMenu: evt ]. "otherwise, explicitly ignore the event if we're not the world, so that we could be picked up if need be" self isWorldMorph ifFalse:[evt wasHandled: false]. ! !

4 4

Re: Cincom Open Sources 'Frost'
by Les Tyrrell 20 Dec '00

20 Dec '00

> > > -----Original Message----- > > From: Eric Arseneau > > Sent: Friday, December 15, 2000 3:08 PM > > Subject: Frost & Pocket Java > > Here is something I was going to send a few weeks back but I never got a > chance to do it. Maybe we can combine the Pocket Java stuff and Frost, as > they do some of the same basic things: > > I have had Java running inside of Pocket Smalltalk for about 2 years now, > but have not had the chance to go very far. In the last few months I have > ported the work to Squeak. Pocket Java now runs under Sqeak. This is also great news- I'm starting to think that we have a viable Camp project here. I looked at your slides and have put in a link on the Frost Wiki under "What is Frost?". I'd like to have see what opportunities exist for collaboration- for instance, I have a hunch that Squeak's 2D and 3D graphical capabilities would be useful in implementing the corresponding Java API's. - les

3 2

← Newer
1
...
4
5
6
7
8
9
10
...
28
Older →

Jump to page: