Squeak-dev November 2001

squeak-dev@lists.squeakfoundation.org

174 participants
374 discussions

A survey of OSs used by Squeakers
by John Hinsley 30 Nov '01

30 Nov '01

Just a little survey. It won't solve anything (the best arguments are those carried on regardless of the facts!) but may tell us something about Squeakind. For me it's a way of trying to resolve Mark's statement that most of his students use Linux and Maarten's assertion that 90% of Squeakers use Windows with my own gut feeling that Mac users are very well represented on this list and that while Windows users outnumber Linux users, it's not by much. Of course Squeak is an OS...... Anyway, anyone who wants to take part in this non-definitive survey, please reply to this *_off list_* (to me, that is) just putting the letter to correspond to the OS they predominently use to run Squeak in the body of the message. And I'll try to collate the results by this time next week. a. Mac b. OSX c. BSD d. Linux c. Other *nix d. Acorn (hi Tim!) e. Windows f. OS2 g. PS2 If you feel that your OS/platform of choice has been cruely neglected, just tell me what you're using. Cheers John -- Reputed to be the reason Windows 2000 was nearly a year late, (paid in shares M$Ds needed the cash and kudos) Netproject's Eddie Bleasdale has renewed his challenge to virus writers. The first person to infect his Linux box wins 10,000 pounds. http://www.silicon.com/bin/bladerunner?30REQEVENT=&REQAUTH=21046&14001REQSU…

10 12

Preliminary Survey Results
by John Hinsley 30 Nov '01

30 Nov '01

The survey generated about 100 mails (a few of which were off topic and a few of which were duplicates due to silly me making a pig's ear of the initial posting). 88 were non duplicated, on topic responses. Some people answered for others, some people answered with more than one OS and either gave no priority or were unable to. Ned Konz, who wins the award for Most Cautious Squeaker presented the most difficult response in that Ned runs Squeak in Linux and Windows, but the Windows version is run within VMware in Linux! In the end I thought I'd be kind to Redmond and count this as Linux and Windows. A preliminary result, by OS a. Mac ////////////////////////////// 30 b. OSX //////////////// 16 c. BSD /// 03 d. Linux //////////////////////////////// 32 e. Other *nix /////////// 11 f. Acorn / 01 g. Windows ///////////////////////////// 29 h. OS2 /// 03 i. PS2 00 WinCE /// 03 Zaurus / 01 Squeak NOS / 01 The Windows category includes 9* and NT Windows. The decision to put WinCE in a different category was pretty arbitary. There's the possibility of some "polution" between the "other *nix" and Windows columns. (Due to stupid error by me.) I'm shocked by the popularity of Mac OSs amoungst Squeakers (I mean, as a statistic, that's *really* contrary to popular wisdom). Otherwise, the "equality" of the main OSs is pretty striking. I'll try and do some other stuff with this, bit, in the meantime, thanks to everyone for their input. Cheers John -- Reputed to be the reason Windows 2000 was nearly a year late, (paid in shares M$Ds needed the cash and kudos) Netproject's Eddie Bleasdale has renewed his challenge to virus writers. The first person to infect his Linux box wins 10,000 pounds. http://www.silicon.com/bin/bladerunner?30REQEVENT=&REQAUTH=21046&14001REQSU…

7 6

Re: Bug tracking
by goran.hultgren＠bluefish.se 30 Nov '01

30 Nov '01

Kevin Fisher <kgf(a)golden.net> wrote: > On Wed, Nov 14, 2001 at 08:55:28PM +0100, goran.hultgren(a)bluefish.se wrote: > > Don't forget JitterBug. It is very simple, only demands Apache as a > > prereq, is easily > > hackable (I have added a bunch nice features to it) uses email and web > > in a neat way. > > Sounds neat! Is this Squeak-based? I've never heard of JitterBug... http://samba.anu.edu.au/jitterbug/ In short: - It is NOT written in Squeak - It is originally written by Andrew Tridgell (The SAMBA guy) - It is written as a few cgi-bin binaries in plain C (but neat and simple code) and thus only needs a normal Apache installation on a Un*x box - It stores the reports as simple files and uses grep for searching - It is VERY easy to add new properties to the search form and the bugreports - You submit bug reports using a simple HTML form. We could easily whip up a simple UI from within Squeak for that too I guess. - It uses email for feedback to the bugreporter and it also receives followup email from the bugreporter and includes that into the system, example with feedback and followup: http://samba.anu.edu.au/cgi-bin/jitterbug/answered?id=84;user=guest - You have "folders" where you can group the reports and you have "users" in the system which can take responsibility of reports. It also has email notification on reports and folders and has good logging so that you can see the history of a report. I just tried it with Scamper and even if the layout looks a bit freaky it seems to work (we can always change the layout to make it look more Squeakish, perhaps mimicking colors etc from the Squeak Swiki) I have added (at my customer) a bunch of new properties (that are searchable to the reports) and also an attachment feature so that you can attach screenshots etc to your report using HTTP POST. I also translated the whole thing into swedish and changed the look a bit - not hard to work with the HTML either. As I said earlier - I am offering to set one up for try-out when Doug Way gives me the word. :-) I have hosts and bandwidth for the try-out, no problem. And then if we like it we move it to Squeakfoundation.org. Ok, now I will shut up. :-) > Well, testing is near and dear to my heart (it's my "real world" job) > so I've always wondered how I could bring Squeak into the game. A nice > Squeak-based defect tracker would be great...the portability would come > in handy as I've always run into problems when the upper crust decides we > need to switch to a new platform every once in a while. :) Plus, integrating > it with a SWiki just seems like the "natural" thing to do.. Well, I actually was not proposing to build a bugreporting system in Squeak - quite the opposite. :-) I just think that setting up Jitterbug (which is robust, works and is hackable) which would take a few hours of my time, is a very reasonable first "go" at a simple system. And then, if we decide to build the super-duper-Squeak-system, we can easily migrate the existing reports into that since it's just files. <CSOTD> "Ok, a little tip then. Perhaps you have seen this in the code? It is a Squeakish way of typing 'PENDING'. #flag: doesn't do anything but this enables us to search for our PENDINGS. Try searching for senders of workAround (by selecting that word and typing Alt-n (Win32) and you will see what I mean). Andreas is obviously fond of this technique! :-)" self flag: #workAround </CSOTD> regards, Göran

4 3

[BUG FIX] JPEGReadWriter2 update
by jmvuletich＠uolsinectis.com.ar 30 Nov '01

30 Nov '01

Hi all "Change Set: JPEGPlugin2 update Date: 30 November 2001 Author: Juan Manuel Vuletich Adds simple dithering when decompressing into a 16bit form. Also avoids generating Color transparent instead of Color black. Includes the needed C support code." For changes to take effect, it's necessary to recompile the plugin. Regards, Juan Manuel Vuletich

1 0

Re: [BUG]Missing method in FileStream
by Bruce ONeel 30 Nov '01

30 Nov '01

Hi, This is done. Thanks Henrik! cheers bruce Henrik Gedenryd <h.gedenryd(a)open.ac.uk> wrote: > Joseph Frippiat wrote: > > > It seems that the method "fileIntoNewChangeSet" is missing in the class > > FileStream. > > > > I get the message window "There may be a problem with your sources file! > > ..." > > > > I use the image "Squeak3.2a-4554" just downloaded from the Internet. > > > > Joseph Frippiat > > > > It's quite ironic that this method with the corrupted source was part of the > very same change set that fixes this corruption bug (see change set 4474). > The source for it got corrupted during the last run of the old version of > that method, which was then replaced by this same change set. > > If you download Squeak3.2a-4566.zip, the problem is fixed and it shouldn't > happen again. If Bruce could remove earlier 3.2a files as well then this > won't happen to others. > > Henrik

1 0

"Packaging" icons in change sets
by Eric Scharff 30 Nov '01

30 Nov '01

I have a bunch of tiny pictures that I use as icons in a UI. Originally I made the UI in Morphic using direct manipulation (filing in the GIF icons by hand) but have encapsulated that stuff into code. What is the best way to get the whole thing into a change set? I would guess the best thing would be to create class methods to generate the Forms, either from the original GIF data or some other form. Is there an easy (or generally accepted) way to do this? Thanks, -Eric

2 1

[BUG]Missing method in FileStream
by Joseph Frippiat 30 Nov '01

30 Nov '01

It seems that the method "fileIntoNewChangeSet" is missing in the class FileStream. I get the message window "There may be a problem with your sources file! ..." I use the image "Squeak3.2a-4554" just downloaded from the Internet. Joseph Frippiat

2 1

[ENH] Hook up non-transforming pretty-printing
by Doug Way 30 Nov '01

30 Nov '01

Here's a changeset which hooks up Henrik's non-transforming parser changes so that the pretty printer and tile scriptor use them. Before filing this in, you'll need to file in Henrik's changesets from Monday. See: http://groups.yahoo.com/group/squeak/message/36671 This new pretty-printing seems to clean up most of the transformation problems. I tested pretty-printing in the browsers, debugger, with diffs, and with tiles. (To turn on pretty printing, use the shift-menu in the browser code pane, or set the preference, or the 'what to show' item, or... ;-) ) Here's an example method which shows the differences in how the pretty-printer worked before and after: myMethod "one" | hey | "two" hey _ hey ifTrue: [nil] ifFalse: ['second']. "three" ^ hey ifFalse: ['first'] ifTrue: ['second']. Previously, the first ifTrue: clause would dissappear completely, and the second ifFalse:ifTrue: would get switched. Now this doesn't happen, although I notice that the 'nil' in the first block dissappears... not sure if this is a buglet. Also, another general oddity is that the "two" comment moves before the temp var declarations (which also happened before). - Doug Way dway(a)riskmetrics.com

1 0

[updates] 14 for 3.2alpha
by Dan Ingalls 30 Nov '01

30 Nov '01

Folks - Herewith a number of bug fixes and enhancements now out on the servers. Enjoy - Dan ----------------------- 4553PolyFixes-ar -- Andreas Raab -- 26 November 2001 Fixes two problems (reported by Kim) when using border styles with polygons: * gribblies when changing the border width * division by zero for certain border widths 4554JPEG2Fixes-ar -- Andreas Raab -- 27 November 2001 Fixes a problem with JPEG2Plugin - if not present you get #primitiveFailures rather than using the default plugin. Also optimizes various bits of code, removing (unneeded) argument tests against the receiver (which would have pointlessly prevented renaming JPEGReadWriter2 into something else). WARNING: The changes (in particular the first fix) require a new plugin to be build." 4555noAudioDir-mir -- Michael Rueger -- 27 November 2001 The 'audio' dir for ChatNotes is now only created if really needed." 4556avoidThumbnailing-sw -- Scott Wallace -- 27 November 2001 Fixes a bug in the mechanism provided for avoiding thumbnailing of certain morphs for parts-bin purposes." 4557haloMenu-sw -- Scott Wallace -- 27 November 2001 An attempt at rationalization of the now hopelessly-unwieldy red-halo menu for morphs. * the clustering together of most (all?) of the standard checkbox-bearing items * the factoring of groups of items into separate methods which return coherent submenus. * menu items mirroring halo actions are provided generically; this allowed deletion of a number of custom methods that had served only to offer halo-action functionality. * full access to all the new border styles is now available via menu. * a new checkbox item governing whether an object should resist removal is added. When set, user confirmation is required before a dismiss-via-halo is carried out. * a few minor rewordings tried out (those in love with 'sticky' might scream.) * a slight regrouping of the world menu, and a fresh factoring of the world's halo menu. * balloon-help provided for all top-level items in standard world menu. * the old #addAddHand... protocol is left behind (though retained for backward compatibility with client code) -- it was *never* used for its announced purpose, and thus only added confusion. All morph-specific additions are now channeled through #addCustomMenuItems... * if the new #selectiveHalos preference is true, then inappropriate halo handles are no longer offered in a halo. Thus, for example, if a morph is marked as being resistant to removal, it will not offer pink (dismiss) handle. NOTE: to regain the former comfort of the old regime, set the selectiveHalos preference back to false. This update sets it to true so you can immediately feel the difference." Preferences addPreference: #selectiveHalos categories: #( Halos) default: true balloonHelp: 'When this is set to true, halo handles that are deemed inappropriate are not shown. Thus, for example, if a morph is marked as resistant to removal, it will not offer a pink ''dismiss''.'. 4558FixPolygonScale-di -- Dan Ingalls -- 28 November 2001 Polygons and related shapes (Curves, Stars) do not need a flex shell to change their size and rotate. However, the Player logic that controls scaling and rotation expects a flex shell to be used. This caused walkbacks in the past, and has been worked around recently by no-oping these operations. This changeSet causes a flex shell to be used when players are involved, but continues the non-flexed control from the halos, owing to its better visual results and consistency with constant border widths when resizing, eg, ellipses and rectangles from the halo. 4559MinesTweaks-di -- Dan Ingalls -- 26 November 2001 Two changes specific to Mines, and one generic to Morphic that greatly speed up creation of the board. 4560noSelectiveHalos-sw -- Scott Wallace -- 29 November 2001 Resets the experimental selectiveHalos preference to false. Relaxes the check for whether a resize handle is wanted." 4561haloCatOops-sw -- Scott Wallace -- 29 November 2001 Fixes a capitalization error in the category of a recently added preference. You'll need to get a fresh Preferences Panel to feel the difference. Change set consists of this preamble only" Preferences setCategoryList: #(halos) forPreferenceNamed: #selectiveHalos 4562GenieUpdateToV202 -- Nathanael Scharli and Stephan Rudlof -- 15 August 2001 *** Genie v2.02 (Update from Genie 1.0) *** This changeset updates Genie to version 2.02. The new features are: - More than 10 times faster lookup (with VM plugin) - New VM plugin (by Stephan Rudlof) and optimized image level code - Execute 'CRRecognizer checkPluginVersion' and watch the Transcript output in order to check availability/compatibility of the VM plugin. - Faster graphical feedback (especially imporant for PDAs) - New dictionary format - Old dictionaries are automatically converted to the new format. This convertion includes modification of the parameters 'speed' and 'curvature relevance'. This is necessary to retain the behavior of the dictionaries with the optimized algorithms. - The new format is backwards compatible. This means that new dictionaries can still be used with Genie v1.0. However, the parameters 'speed' and 'curveture relevance' are interpreted slightly different, which may lead to slower recognition or inaccuracies in Genie v1.0. - Improved UI (Improved morphs to inspect/modify gesture dictionaries resp. display properties) - BugFixes IMPORTANT information about the Genie VM plugin The Genie VM plugin makes Genie about 10 to 20 times faster. Especially on slower machines (PDA, etc.), it is really worth using it and it makes Genie much more fun!! Testing the plugin To see whether the plugin is available and compatible to the image level code execute 'CRRecognizer checkPluginVersion' and watch the Transcript output. To compare the performance of the primitive to the original performance do the following: 1) Load the gesture dictionary 'http://www.iam.unibe.ch/~schaerli/genie2/GenieTest.ggd' in your image. 2) Execute 'CRStrokeFeature testPrimitivePerformance' and watch the transcript output. 3) Execute 'CRStrokeFeature testOriginalPerformance' and watch the transcript output. 4) Compare the output of steps 2) and 3). The primitive should be about 50 times faster. If you are not sure whether the primitive works correctly, do the following: 1) Load the gesture dictionary 'http://www.iam.unibe.ch/~schaerli/genie2/GenieTest.ggd' in your image. 2) Execute 'CRStrokeFeature testPrimitive' and watch the transcript output. Compiling and installing the plugin The Genie plugin may already be part of your VM (execute 'CRRecognizer checkPluginVersion' to check). If not, you can compile and install it yourself, by doing the following: 1) Generate the C-file by executing 'GeniePlugin translate'. This writes the file 'GeniePlugin.c' into the subdiretory 'GeniePlugin'. 2) Compile the plugin and make the resulting dynamic library available for the VM. (This step is platform dependent. On Windows, you have to get the VM sources, build the plugin (enter 'build GeniePlugin.dll') and copy the resulting file GeniePlugin.dll into the Squeak VM folder). --> For more information about Genie, see the class comment of the class AGenieIntroduction." 4563DirectionHandles-ar -- Andreas Raab -- 29 November 2001 Makes the direction handles option a per-morph on/off preference. Only stores its value if the morph goes against the system wide pref." 4564arrowDeltaEtc-sw -- Scott Wallace -- 29 November 2001 Allows the deltas applied via little green arrows on numeric readout tiles to be specified (formerly, the delta had always been one) -- and uses this to make the stepsize of the arrows for #scaleFactor be .1. Also makes the default float precision of #scaleFactor be such that one decimal place is always shown." 4565FixPolygonHeading-di -- Dan Ingalls -- 29 November 2001 Further to the changes of 'FixPolygonScale', this changeSet (by removing a method ;-) causes heading to be reported correctly for both flexed and unflexed polygons. 4566TextSuccessorFix-di -- Dan Ingalls -- 29 November 2001 Kim Rose reported a while ago that there was a walkback when trying to turn a GeeMail from 1-column book to a 2-column book and Leo Burd recently noted the same problem. Bob Arning reported a related problem with simple TextMorphs: In a new TextMorph choose 'add successor' from its menu Try to add text to first TM, get 'Error: Inconsistency in text editor' This change fixes the simple manifestation of this problem, and hopefully the GeeMail problem as well, though I have not tested it.

1 0

Re: An Ideal System Browser
by Richard A. O'Keefe 30 Nov '01

30 Nov '01

Jerry Balzano <gjbalzano(a)ucsd.edu> wrote: Not that I'd presume to know what that is at my stage of (Squeak) development, but it occurred to me as I was browsing around through various classes and instances and methods that the Squeak System Browser lacks a feature that we take for granted on all our web browsers and that most of us would be lost without -- a "back" button! Is it possible that nobody has found the need for such a function? 'Cause I'd be looking at a method and I'd find a word I didn't understand or some such, or just say "I wonder about" some other class, method, or whatnot, and go there, then say to myself, "OK, that's great, (or "that's complicated" or whatever) now I'm ready to resume what I was doing." But if what I was doing was in some totally separate class with lots of methods to paw (OK, scroll) through, well you get the idea -- what a pain! and how wasteful of my time and energy. I can't think what I'd do with a "back" button. Every time I visit something new, I get a new window. All my recent viewing history is right there visible on the screen. I see a class name I'd like to follow up, I select it, I type Cmd-b, and I get an additional window. I see a method name, and do Cmd-n or Cmd-m, and I get an additional window. To make assurance doubly sure, in Morphic, I can duplicate things using the [][] green icon at the top right of the halo; that way I can copy an entire browser window. In fact, I tend to end up with way too many windows all over the place, and periodically have to go around closing them. How does one use Squeak so that a "back" button would be useful?

1 0

← Newer
1
2
3
4
5
6
7
...
38
Older →

Jump to page: