Squeak-dev October 2004

squeak-dev@lists.squeakfoundation.org

156 participants
313 discussions

Re: [FIX][BUG] Float NaN's ( retracting patch for #testNaN2 )
by Richard A. O'Keefe 02 Oct '04

02 Oct '04

"Boris Gaertner" <Boris.Gaertner(a)gmx.net> wrote: I use Squeak in Windows, with an Intel Celeron. On that platform the expressions 1.0e200 sin and 1.0e200 tan answer two different representations of NaN. which is certainly a wrong answer. The Intel documentation (IA-32 Intel Architecture Software Developer's Manual) says for instructions FSIN and FPTAN: <citation> The source operand must be given in radians and must be within the range -2^63 to +2^63. </citation> which is why a C compiler *MUST NOT* convert calls to sin() and tan() directly to uses of the FSIN and FPTAN instructions, unless the programmer has explicitly requested a non-standard fp mode (rather like the "-fnonstd" option of the Sun C compiler or the "-ffast-math" option of GCC. Source values outside the range -2^63 to +2^63 can be reduced to the range of the instruction by subtracting an appropriate integer multiple of 2*pi or by using the FPREM instruction with a divisor of 2*pi. See the section titled 'Pi' in Chapter 8 of the IA-32 Intel Architecture Software Developer's Manual, Volume 1, for a discussion of the proper value to use for pi in performing such reductions. Which is what someone implementing sin() and tan() in the maths library might do, yes. Although it is not the best advice. The Sun math library uses "infinite precision pi" for argument reduction. This can make testing a bit tricky as sin(3.14.......) is non-zero in the Sun library even for the representable number closest to pi. Demonstration: f% cat >sin.c #include "math.h" int main(void) { printf("%g\n", sin(M_PI)); return 0; } <EOF> f% cc sin.c -lm f% a.out 1.22465e-16 Where this _really_ matters of course is precisely for those very large argument values; using "infinite precision pi" and using the Intel recommended value will give you quite different answers. In other words: The implementor of a maths library has the permission to do whatever he thinks is suitable. No, that's not what it says. They cannot give that permission. Permission to deliver bad answers in the name of speed can only come from the language standard or the programmer who invokes the compiler.

3 3

Call for help [was: [Re: [ENH] BehaviorHashEnh]]
by Stephan Rudlof 02 Oct '04

02 Oct '04

Stef, I'm trying to think freshly, since this is very weird: I have measured with a fresh 3.8alpha... My results for the IdentitySet variant have been *much* worser than yours, with or without the cs (which is expected, since the behavior of IdentitySet won't be affected by the cs). For this reason, the idea of the cs has been to introduce Behavior>>hash - using String>>hash - to avoid the default Object>>hash which itself calls ProtoObject>>identityHash. This is also the reason, why the IdentitySet variant remains worse for my system (it just continues to use >>identityHash). But the cs doesn't seem to have a relevant effect in your system, because *your* >>identityHash seems to work much better than mine! Now the question: why? ProtoObject>>identityHash calls a prim: which VM are you using? It would be nice, if some other person would doIt the following expression and post the result together with the VM version (squeak -version) and processor version to get some more data. DoIt ---- | allClasses allClassesSet block | allClasses := Smalltalk allClasses. block _ [allClassesSet _ allClasses asIdentitySet. [allClasses do: [:class | allClassesSet remove: class]] timeToRun]. {block value. block value. block value}. For me the result is #(3112 3166 3108) for sr@karl:~/DepS/Tmp$ squeak -version 3.7b-5 #1 Mon Jun 14 18:05:35 CEST 2004 gcc 3.3.3 Squeak3.7beta of '1 April 2004' [latest update: #5954] Linux karl 2.4.26sr #1 Sun May 23 19:28:24 CEST 2004 i686 GNU/Linux default plugin location: /usr/local/lib/squeak/3.7b-5/*.so at a 366MHz Pentium. Any additional ideas for the huge difference between mine and Stefs results are also welcome. Greetings Stephan stéphane ducasse wrote: > | allClasses allClassesSet block | > allClasses := Smalltalk allClasses. > block _ [allClassesSet _ allClasses asSet. > [allClasses do: [:class | allClassesSet remove: class]] timeToRun]. > {block value. block value. block value} > > one before > #(27 27 25) > one after > #(31 35 32) > > | allClasses allClassesSet block | > allClasses := Smalltalk allClasses. > block _ [allClassesSet _ allClasses asIdentitySet. > [allClasses do: [:class | allClassesSet remove: class]] timeToRun]. > {block value. block value. block value}. > > two before > #(19 19 19) > two after > #(37 19 22) > > > I hope this helps. This was with 6273 > > Stef > > -- Stephan Rudlof (sr(a)evolgo.de) "Genius doesn't work on an assembly line basis. You can't simply say, 'Today I will be brilliant.'" -- Kirk, "The Ultimate Computer", stardate 4731.3

1 3

Last Call for Participation
by Jeff Eastman 01 Oct '04

01 Oct '04

We have gotten a fair number of responses to our Smalltalk Survey (http://www.windwardsolutions.com/smalltalksurvey.htm) but more would improve our analysis. I won't bug you folks about this any more but remember, only those who have participated in the survey will be able to access the results. I will close out the collection process in a few days, so please take a moment to post a brief response. Thanks in advance, Jeff Eastman

1 0

[BUG]Debugger error:MNU:undefinedObject>>firstSubmorph:
by karl.ramberg＠chello.se 01 Oct '04

01 Oct '04

I got this error and could only do a screen capture of it. I've been noticing that the emergency evaluator is popping up quite often, I have not catched the error/bug though... I attached the emergency evaluator screen shot. Karl

2 2

Re: [ENH] BehaviorHashEnh
by stéphane ducasse 01 Oct '04

01 Oct '04

hi Stefan with 3.8 6273 I get | allClasses allClassesSet block | allClasses := Smalltalk allClasses. block _ [allClassesSet _ allClasses asSet. [allClasses do: [:class | allClassesSet remove: class]] timeToRun]. {block value. block value. block value} #(51 48 53) and nearly the same in 6208 Stef could you check because something certainly changed. On 1 oct. 04, at 04:21, Stephan Rudlof wrote: > Hello Stef, > > nice to see this reviewed! > > For me there is a dramatic improvement with the cs, copied from a > workspace: > > Test: > doIt > | allClasses allClassesSet block | > allClasses := Smalltalk allClasses. > block _ [allClassesSet _ allClasses asSet. > [allClasses do: [:class | allClassesSet remove: class]] timeToRun]. > {block value. block value. block value} > #(7127 7135 7120) #(7102 7161 7140) #(7341 7353 7300) "before fileIn" > #(251 250 255) #(257 255 250) "after fileIn" > > before and after filing in this cs. > To see the problem again just doIt > | allClasses allClassesSet block | > allClasses := Smalltalk allClasses. > block _ [allClassesSet _ allClasses asIdentitySet. > [allClasses do: [:class | allClassesSet remove: class]] timeToRun]. > {block value. block value. block value}. > #(3677 3681 3664) #(3650 3688 3654) #(3656 3695 3648) "before" > #(3662 3686 3663) #(3645 3710 3647) "after" > > > I'm using Squeak3.7#5988. > > ducasse(a)iam.unibe.ch wrote: >> can somebody else have a look at this enh. >> >> I run the "tests" >> >> | allClasses allClassesSet block | >> allClasses := Smalltalk allClasses. >> block _ [allClassesSet _ allClasses asSet. >> [allClasses do: [:class | allClassesSet remove: class]] timeToRun]. >> {block value. block value. block value} >> >> >> before and after filing in this cs. >> To see the problem again just doIt >> | allClasses allClassesSet block | >> allClasses := Smalltalk allClasses. >> block _ [allClassesSet _ allClasses asIdentitySet. >> [allClasses do: [:class | allClassesSet remove: class]] timeToRun]. >> {block value. block value. block value}. >> >> >> after loading I got >> -> #(64 57 63) >> >> To see the problem again just doIt >> -> #(34 33 39) >> >> > >> So I'm confused because I thought this would be the opposite so I may >> be >> too tired. > > For me it is the opposite: e.g. 251 to 3662! > > 1. Observations: > - My numbers for the first doIt after fileIn (trying to show the > improvement) seem to correspond to your numbers of *both* doIts in > magnitude, if you have a GHz machine (I have a slow 366MHz Pentium). > - The proportion of your numbers *after* fileIn is similar to the > proportion of mine *before*, but then mine are *much* slower. > > 2. Since you have been tired: the doIts differ, and the cs comment of > v1.1 says *not* to fileOut the cs to fileIn it later again. > > 3. Questions: > - What have you gotten before fileIn for both doIts? > - What is the previous version stamp of String>>hash in your test > image? For me it is SqR 8/13/2002 10:52. > > 4. Thoughts: > - Has there been a change of SystemDictionary>>allClasses? For me it > returns the classes and not their names (symbols). > - Not very probable: Do you have a better identityHash? > > My VM: > squeak -version > 3.7b-5 #1 Mon Jun 14 18:05:35 CEST 2004 gcc 3.3.3 > Squeak3.7beta of '1 April 2004' [latest update: #5954] > Linux karl 2.4.26sr #1 Sun May 23 19:28:24 CEST 2004 i686 GNU/Linux > default plugin location: /usr/local/lib/squeak/3.7b-5/*.so > > > Hope that helps, > Stephan > >> >> Stef >> > > -- > Stephan Rudlof (sr(a)evolgo.de) > "Genius doesn't work on an assembly line basis. > You can't simply say, 'Today I will be brilliant.'" > -- Kirk, "The Ultimate Computer", stardate 4731.3 >

2 1

Chemistry classes?
by Daniel B. Faken 01 Oct '04

01 Oct '04

Hello Fellow Squeakers, I'm working on some molecular-visualization stuff for Croquet. Does anyone have/know-of classes for representing Atoms/Molecules in Squeak/Smalltalk? I don't so much need something to render them (though that would be nice) as to represent them (so that I can access their coordinates, bonds, elemental properties, etc.) & manipulate them. I've looked at the 'Atomic' game and the 'Bouncing Atoms' sim, but neither of those provides any real support. Also, is anyone using Squeak for teaching Chemistry? thanks for any help! Daniel Faken

2 1

[Q] fastest way to get polygons on the screen
by Baveco, Hans 01 Oct '04

01 Oct '04

Hi all, I am working on the drawing part of my shape-file reading package, and wondered what would be the fastest way to get all these polygons on the screen. Being geographically referenced, the vertices need to be scaled and transformed. I did this by first creating a Form of the desired extent, than defining a BalloonCanvas on this Form, and calculating a scaling factor. form _ Form extent: aPoint depth: anInteger. polyColor _ self polygonColor. arcColor _ self arcColor. form fillColor: self backgroundColor. canvas _ BalloonCanvas on: form. canvas aaLevel: 4. geoBB _ self boundingBox. scaling _ form extent x / geoBB extent x min: form extent y / geoBB extent y. Next with this scaling factor I create a MatrixTransform2x3, attach this to the canvas, and iterate over all polygons, using the #drawGeneralPolygon:color:borderWidth:borderColor: method to do the display. m _ MatrixTransform2x3 withScale: 1 @ -1 * scaling. m offset: self xMin negated @ self yMax * scaling. canvas transformBy: m. 1 to: self numberOfShapes do: [:i | | contours | contours _ self verticesByPartsForShapeIndex: i. canvas drawGeneralPolygon: contours color: "polyColor" Color random borderWidth: 1 borderColor: arcColor] Is this the approach with the best possible performance, given that scaling and transformation will always be necessary? Would it be advisable to cache scaled and transformed polygons, or is scaling and transformation within the #drawGeneralPolygon:color:borderWidth:borderColor: method not too costly? I would like to reach the same display speed for a shape file as the ESRI ArcView GIS (though they may use all kinds of smart optimizations there ...) ;-) Comments and advise are really appreciated! Hans

2 1

3D acceleration.
by Alan Grimes 01 Oct '04

01 Oct '04

Now that I finally have some level of hardware 3D acceleration working at all on my linux computer, Istarted to look at wheather Squeak is able to access it. Attempting to toggle hardware acceleration with the menu checkbox was ineffectual. Now I have these questions: 1. On what platforms is hardware 3D acceleration supported? 2. What things need to be done to make it work on linux? My actual motovation for poking around with Squeak's 3D facilities comes from my good old AI project. The development plan which I am in the process of writing calls for a 3D environment with a very good facimilie of a physicality, implementing all of newtonian physics if not some parts of the laws of thermodynamics. The present implementation appears to be purely a rendering engine. Does anyone have any suggestions on where I should go to learn about how to add physical simulation features to the engine? I had attempted to make a posting that included some comments about a possible VM enhancement from the image itself with mixed results. I'll repeat them here because I don't think many people were able to read the earlier posting. In working on the VM, I noticed that it actually had 3 stacks instead of the two mentioned to me earlier. These include the parameters stack, the context stack, and the hardware stack. While a fair ammount of effort has gone into minimizing the use of the hardware stack, for good reason apparently, It _MIGHT_ be advantageous to attempt to combine either of the two software-managed stacks with the hardware stack for the sake of conserving registers and reducing the number of data items that need to be maintained explicitly by the VM code. I was wondering if anyone had done any work along these lines or may be able to point me to an argument that shows the current design to be superior.

4 9

[UPDATE] please check and report problems :)
by stéphane ducasse 01 Oct '04

01 Oct '04

hi guys I pushed a couple of cs in the unstable stream. - Please check your your fixes are solving the problems you were fixing. - There are some pending cs with a conflicts tag by marcus or me so please have a look (I do not want to solve conflicts nor can nor is willing to push them in unstable in that state). - Please have a look at the enh were I explicit mentioned help. Stef

2 1

Re: [FIX] [IRC]IRC_assureLogDirExistence
by Frank Caggiano 01 Oct '04

01 Oct '04

On Sep 30, 2004, at 14:01, danil-home(a)tsnet.ru wrote: > > Launching IRC in fresh squeak dir gives walkback - because there is no > irc-log directory structure yet. > This tiny fix forces IRCErrorLog class>> logStreamNamed: existsIn: to > create log directory if it doesn't exist. > <IRC_assureLogDirExistence.cs.gz> > I'm not seeing this error when I start up IRCe in a fresh squeak directory. Could you supply more info please. Which version of IRCe, what version of Squeak, how are you loading IRCe into the image, etc. Thanks Frank Caggiano

1 0

Jump to page: