Squeak-dev November 1999

squeak-dev@lists.squeakfoundation.org

164 participants
545 discussions

XJ320
by Torsten.Bergmann＠phaidros.com 30 Nov '99

30 Nov '99

I've found an old PPD power point file from the user conference 1996. Glenn Krasner mentioned a mail adress (x3j20(a)qks.com) for subscribing a discussion group about the ANSI Smalltalk standard. (subsribe x3j20 <firstname> <lastname>) Does anyone know if the work is going on on this list ? -Torsten

1 0

Re: caching of screen display...
by John M McIntosh 30 Nov '99

30 Nov '99

> A week back I asked about the poor performance of X-Windows updates. > > I was told lots of bits flow across the wire. > > So I looked > PPS stXPutImage looks like an interesting place to update with some caching > code eh? > Well after a few hours of programming I can confirm by deferring updates one can improve visual performance. For example in my x-windows test case 10 screen restores take 78.116 seconds. By deferring updates I can cut this update time to 18.760 seconds. However you visually miss lots of screen updates, but would you rather wait? FYI 100 screen restores on a local screen take 34,233 updates and 41 seconds, however I couldn't wait for that many across ethernet. But I can cut the updates to 1885, a more reasonable number. If I ensure screen updates occur at least every 1/20 of a seconds morphic animations flow at a reasonable speed. In any case I'll look at bundling this all up for interested parties. Although I'm doing this down in sqXWindow.c I'm going to explore moving the deferring logic up into the Smalltalk VM and consider an adaptive algorithm to defer updates based on some yet to be determined heuristic. -- =========================================================================== John M. McIntosh <johnmci(a)smalltalkconsulting.com> 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com =========================================================================== Custom Macintosh programming & various Smalltalk dialects PGP Key: DSS/Diff/46FC3BE6 Fingerprint=B22F 7D67 92B7 5D52 72D7 E94A EE69 2D21 46FC 3BE6 ===========================================================================

1 0

[NEWBIE] Fonts/Sizes in Squeak
by Udo Schneider 30 Nov '99

30 Nov '99

a newbie question: How do I create a Text with different Fonts/Sizes in Squeak. I found classes for TextColor, TextKern etc. but nothing for TextSize or TextFont. Did I miss something? Thanks, Udo

1 0

RE: Intervals
by David N. Smith (IBM) 30 Nov '99

30 Nov '99

At 13:14 -0500 11/29/99, Jarvis, Robert P. wrote: >The attached change set combines Andrew's changes with Dan's suggestions and >runs correctly for the test cases identified, as well as for negative >numbers, i.e. Hummmm... I've been looking at the code and I wonder if it will always work with floating point values? Stepping through the interval involves adding the step value at each increment, but asking if a value is in the interval involves modulo arithmetic. The problem comes when the step value cannot be exactly represented in binary (0.7 for example). I can't believe it will always work identically, but I've not been able to come up with a test case that fails. (I implemented a version if #includes: that is simply: includesP: aValue ^ self asArray includes: aValue and used it for comparison. This IS a precise and accurate solution, though it might run out of memory at inconvenient times and otherwise misbehave when the interval has lots of elements. Dave _______________________________ David N. Smith IBM T J Watson Research Center Hawthorne, NY _______________________________ Any opinions or recommendations herein are those of the author and not of his employer.

3 2

RE: Intervals
by Andrew P. Black 30 Nov '99

30 Nov '99

>Shouldn't that test on step compare it with ZERO? It fails miserably >when the step is 0.5. > >includes: aNumber > | offset | > step = 1 ifTrue: [^ aNumber between: self first and: self last]. > step > 0 > ifTrue: [offset _ aNumber - start. > ^ (aNumber between: start and: self ... Yes, David is right. (embarassed smile) And I subjected this to extensive random testing, too -- but only for integer step sizes :-( Andrew

1 0

Re: Intervals
by Leandro Caniglia 30 Nov '99

30 Nov '99

>So when's that type inferencing engine going to be released, Dan? :-) Notably I'm not Dan, but let me tell you that some guys at mathMorphs are working on a type inference system for Squeak. We have followed the Cartesian Product Algorithm by Ole Agensen. In this implementation we have introduced some "extensions" to the original CPA in order to handle instance variables and work around other inconveniences. We are still working on but want to invite everybody who is willing to criticize or contribute to this interesting project. We hope to set up soon a new mathMorphs Swiki with a TI section open to discussion and testing. Please, don't take this mail as an announcement, but as a simple comment on this recurrent and important topic. /Leandro

1 0

Re: Intervals
by merlyn＠stonehenge.com 30 Nov '99

30 Nov '99

>>>>> "David" == David N Smith $IBM$ <David> writes: David> I can't believe it will always work identically, but I've not been David> able to come up with a test case that fails. (I implemented a version David> if #includes: that is simply: David> includesP: aValue David> ^ self asArray includes: aValue David> and used it for comparison. This IS a precise and accurate solution, David> though it might run out of memory at inconvenient times and otherwise David> misbehave when the interval has lots of elements. You could even do this: includesP: aValue self do: [:each | each = aValue ifTrue: [^true]]. ^false That won't ever run out of memory. Unless you got one of those open-ended Interval's. :) Hey wait, that's the definition in Collection! Geez, didn't I say this a few months ago? Interval|includes: is broken, and should simply be removed. If you want an optimization for integer step classes, be sure to call super|includes: when it's not one of those. Otherwise, just delete the message entirely, and let it fall from Collection. It all comes full circle. :) -- 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

Closing in on 2.7
by Dan Ingalls 30 Nov '99

30 Nov '99

Folks - I want to close 2.7 to changes in the next day or so. I have a couple of Scamper changes from the GA Tech folks, and a few other things, but I'd like to end it there for now. If you have a burning need, or I've left out something that is important, please let me know offline. On top of all this, I hope then to put out Squeak Name Spaces v1, and Ian's J3 changes, and then spend a few days testing the whole assembly. As soon as it looks stable, we'll call it soup and write a new set of archives (rats -- we still don't have the Squeak decompressing non-line-feeding, one-file-runs-on-all-platforms self-extracting release package). I'm expecting neither SNS nor J3 to be fully operational at the time of the 2.7 release (but then, who knows ;-). I just want to get most of the changes in, and the rest of the system stable. Then we will have things in place for a lot of people to help test SNS and J3 going forward without having to absorb a lot of changes. I'm expecting that 2.7 (besides being an ever-better Squeak of course) will mainly be a beta testing platform for SNS, J3, segment swapping (which becomes more interesting with SNS), and a nice foreign function interface that Andreas and others are working on. That's the plan, anyway. - Dan, for SqC

2 1

[UPDATES] 11 more for 2.7alpha
by Dan Ingalls 30 Nov '99

30 Nov '99

Folks - Herewith a number of changes, the bulk being four files relating to Forms in Scamper. I fired up Scamper and observed some problems, so we may have garbled things a bit, or it could be an interaction with Worlds-in-Worlds that get used now if you open Scamper in an MVC world. Anyway, please try it, and I'll turn around any necessary fixes ASAP. Also note that AlignmentTweaks makes changes to 4 central methods in AlignmentMorph. I have tested these, but there may be repercussions in other applications. It should at least be faster. Enjoy - Dan -------------------------- 1661veryDeepCopy-hg -- Henrik Gedenryd with tweaks by Ted Kaehler -- 24 November 1999 When copying many small structures, the variable check's 'once-in-a-while' becomes way too often, so I changed this to use time intervals instead, since the need is related to the programmer changing the code. Ted set the time interval to 10 minutes. The idea is to flag it while the programmer is still looking at it. Also, I've given each class a chance to set the initial size of the array, so that when an object of class X is called, if you know that as a rule of thumb it contains just a small object hierarchy or uses many shallow copies, you can set a smaller size. (It will always grow if needed.) This one alone gave a 400% speedup for a very small structure. Beside the speedups, I have also added copying of dependents to veryDeepCopy. (Only applies to MVC. Morphic has its own dependents inst var.) This slows down the copying a bit, esp. checking for 'backward' dependencies, so that objects that have the source as a dependent also gets the copy as one (unless the source is copied too; then that copy gets it). The whole DependentsFields structure is scanned once per group of objects copied. Also fixed veryDeepCopy of a live Browser. Also fixed Float.closeTo: for NaN and Infinity." 1662InterpVersion-ar -- Andreas Raab -- 27 November 1999 Make the version accessible from which the interpreter was generated. NOTE TO VM IMPLEMENTORS: The system attribute 1004 has been assigned to return the VM version. A string named 'interpreterVersion' is now being exported by sq.h and should be returned upon queries to Smalltalk>>getSystemAttribute: 1004. " 1663FileListBugFix-sge -- Steve Elkins -- 28 November 1999 Parameterizes 2 very similar FileList methods, #addNewDirectory and #addNewFile, as well as removing the same bug from each, that pushing the cancel button in the FillInTheBlank dialog gave a walkback." 1664HandMorph-changeColor Better positioning of ColorPicker for desktop 1665AlignmentTweaks -- Dan Ingalls -- 25 November 1999 An attempt to stamp out rampant redisplay in Alignment morphs when very little is really changing. 1666FlexDragFix -- Dan Ingalls -- 28 November 1999 Restores flexed dragging that was disbled by update #1596. Also fixes the offset used. 1667ScamperForms-Fixes1 -- Bolot Kerimbaev and Lex Spoon -- 26 November 1999 Minor tweaks, make Scamper form handling more like industry web browsers: - drop down list (enhancement) - select list defaults to size 1 - select list of size 1 uses drop down list - separate display and value options for lists - text area defaults to size 20x2 - text area to use inboard scrollbars - password field support (as regular text) - checkbox now supported - form starts from/ends with a new line" 1668Forms-bf -- Bert Freudenberg -- 4 November 1999 Prerequisites: Forms (Bolot/Lex) buttons-bf (myself) Makes checkboxes and radiobuttons in Scamper forms look like checkboxes and radiobuttons." 1669MIMEDocEnh 1670ScamperForms-Fixes2 -- Bolot Kerimbaev -- 27 November 1999 Some more fixes and enhancements for Scamper: - password font (all chars=$*) - upload form field support password font is generated each time it is used" 1671FreeCell-satisfaction -- David Pennell -- 29 November 1999 change contains: to anySatisfy: due to changes made in update 1619."

1 0

caching of screen display...
by John M McIntosh 29 Nov '99

29 Nov '99

A week back I asked about the poor performance of X-Windows updates. I was told lots of bits flow across the wire. So I looked On the mac in my test case of doing 100 Display restore I get 55,320 calls to ioShowDisplay and of course to the Mac's CopyBits. Nice but expensive. My mac does this in 52 seconds, so I'm getting at least 1000 screen updates a second. I changed the code to cache updates which I then trigger each 1/60 of a second if required. This reduces the CopyBits calls to 2417. Now Before I hack the X-Window's implementation I'm wondering if some other squeaker has taken this path before? PS Lastly I'm actually not caching the incoming bits since I assume we are drawing from special object TheDisplay, and I can just cache the region to display versus all the bits. I couldn't see a case where the incoming bitmap was different, but I am willing to be correct this assumption this early in the morning. PPS stXPutImage looks like an interesting place to update with some caching code eh? -- =========================================================================== John M. McIntosh <johnmci(a)smalltalkconsulting.com> 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com =========================================================================== Custom Macintosh programming & various Smalltalk dialects PGP Key: DSS/Diff/46FC3BE6 Fingerprint=B22F 7D67 92B7 5D52 72D7 E94A EE69 2D21 46FC 3BE6 ===========================================================================

2 1

← Newer
1
2
3
4
5
6
7
...
55
Older →

Jump to page: