November 2010 - Pharo-dev - lists.pharo.org

Re: [Pharo-project] Issue 3247 in pharo: Remodel Text Constants

by pharo＠googlecode.com

Comment #11 on issue 3247 by stephane.ducasse: Remodel Text Constants http://code.google.com/p/pharo/issues/detail?id=3247 Ben when a class uses a Pool you can use the key of the variable instead of the at: expression For example: text: aText "Answer an instance of me such that the text displayed is aText according to the system's default text style." ^self new setText: aText textStyle: DefaultTextStyle copy offset: 0 @ 0 is better than text: aText "Answer an instance of me such that the text displayed is aText according to the system's default text style." ^self new setText: aText textStyle: (SharedDictionary at: #DefaultTextStyle) copy offset: 0 @ 0

Nov. 13, 2010

Re: [Pharo-project] Issue 3247 in pharo: Remodel Text Constants

by pharo＠googlecode.com

Comment #10 on issue 3247 by stephane.ducasse: Remodel Text Constants http://code.google.com/p/pharo/issues/detail?id=3247 I have the impression that we should add a ClassVar to TextStyle and that a lot of changes where simply puts in the wrong place. all the methods in TextStyle class for example. Similarly HostFont should have a classVariables for its fonts! Same probably for StrikeFont May be should perform these changes first. Like that the migration will be easier.

Nov. 13, 2010

Re: [Pharo-project] Issue 3247 in pharo: Remodel Text Constants

by pharo＠googlecode.com

Comment #9 on issue 3247 by stephane.ducasse: Remodel Text Constants http://code.google.com/p/pharo/issues/detail?id=3247 The TextConstant class>>initialize could be better since we cannot reproduce it (or somehow) Ctrll := $ should be something like Ctrll := Character value: XCXXC like that we do not really on an invisible character

Nov. 13, 2010

Re: [Pharo-project] Issue 3247 in pharo: Remodel Text Constants

by pharo＠googlecode.com

Updates: Cc: -siguctua Comment #8 on issue 3247 by stephane.ducasse: Remodel Text Constants http://code.google.com/p/pharo/issues/detail?id=3247 Then I do not understand why You changed | letter varAndValue tempArray width | "CtrlA..CtrlZ, Ctrla..Ctrlz" letter := $A. #( 212 230 228 196 194 226 241 243 214 229 200 217 246 245 216 202 210 239 211 240 197 198 209 215 242 231 1 166 228 132 130 12 232 179 150 165 136 153 182 14 15 138 17 18 19 11 21 134 145 151 178 167 ) do: [:kbd | TextContants at: ('Ctrl', letter asSymbol) asSymbol put: kbd asCharacter. letter := letter == $Z ifTrue: [$a] ifFalse: [(letter asciiValue + 1) asCharacter]]. into | letter varAndValue tempArray width | "CtrlA..CtrlZ, Ctrla..Ctrlz" letter := $A. #( 212 230 228 196 194 226 241 243 214 229 200 217 246 245 216 202 210 239 211 240 197 198 209 215 242 231 1 166 228 132 130 12 232 179 150 165 136 153 182 14 15 138 17 18 19 11 21 134 145 151 178 167 ) do: [:kbd | SharedDictionary at: ('Ctrl', letter asSymbol) asSymbol put: kbd asCharacter. letter := letter == $Z ifTrue: [$a] ifFalse: [(letter asciiValue + 1) asCharacter]]. The idea is to have TextConstant (the new shared) containing the constants and the other dictionary the variable part but here this is clearly constant.

Nov. 13, 2010

Re: [Pharo-project] Issue 3247 in pharo: Remodel Text Constants

by pharo＠googlecode.com

Comment #7 on issue 3247 by stephane.ducasse: Remodel Text Constants http://code.google.com/p/pharo/issues/detail?id=3247 Then we could define such dictionary as a classVar of the new sharedPool class TextConstant. Like that we minimize the globals in the system.

Nov. 13, 2010

Re: [Pharo-project] Issue 3247 in pharo: Remodel Text Constants

by pharo＠googlecode.com

Comment #6 on issue 3247 by stephane.ducasse: Remodel Text Constants http://code.google.com/p/pharo/issues/detail?id=3247 Benjamin, it would be good to rename SharedDictionary into something like TextSharedInformation.

Nov. 13, 2010

Re: [Pharo-project] Issue 3247 in pharo: Remodel Text Constants

by pharo＠googlecode.com

Comment #5 on issue 3247 by stephane.ducasse: Remodel Text Constants http://code.google.com/p/pharo/issues/detail?id=3247 I read all the code and this looks quite good. I will run it now and check if I can release package after.

Nov. 13, 2010

[Pharo-project] Issue 3256 in pharo: matchesRegex on colon

by pharo＠googlecode.com

Status: New Owner: ---- New issue 3256 by davide.v...(a)gmail.com: matchesRegex on colon http://code.google.com/p/pharo/issues/detail?id=3256 Pharo image: Pharo-core Pharo core version: 1.2-12146 Virtual machine used: Squeak 4.2.5beta1U for mac If you evaluate: 'whatever' matchesRegex: ':' it rises an error. if I escape the colon the error disappears, but is it right? I think it's a sort of coupling with method parsing. Tx Davide

Nov. 13, 2010

Re: [Pharo-project] Howto measure memory footprints of functions/objects/libraries in Pharo?

by Alexandre Bergel

Ok, I will check. Alexandre On 12 Nov 2010, at 17:40, Max Leske wrote: > Hi Alexandre > > The weird behaviour seems to be related to inline compilation in workspaces. I moved the execution into a method, now everything's back to normal. > > In the meantime I verified the previously described scenarios with a fresh one-click image (Pharo 1.1). > One really dramatic example is this: > > repo := FSGitRepository on: FSDiskFilesystem current working / 'foo'. > MessageTally spyOn: [ repo tagNames ] > > ---> 3ms for a huge repository with 10 tags. > > > MemoryProfiler > viewProfiling: [ repo tagNames ] > > ---> low space warning (same repository etc.) > > > I hope you might have an idea what's going wrong. > > Cheers, > Max > > > > > Hi again. > Please ignore the last e-mail for the moment. There's something weird going on in my image. I'll run some tests on a fresh official image first and then post back. > Max > > > --------------------------------- > Thanks Alexandre. > Now I tried out your example and they worked perfectly. I have run into two problem though which I don't know how to deal with: > 1. When I run anything else than what you suggested, I do get the window for the profiling view but it is empty. E.g when I run this: > > MemoryProfiler > viewProfiling: [ 1 to: 100000 do: [ :i | > (FSMemoryFilesystem new root / ('foo', i asString)) > writeStreamDo: [ :stream | stream nextPutAll: 'foo' ] ] ] > 2. When I run the code I am actually interested I get a low space warning. Example: > repo := FSGitRepository on: FSMemoryFilesystem new root / 'foo' > MemoryProfiler > viewProfiling: [ repo head ] > > I'll be happy to run any tests you suggest. If you'd like to run the code yourself, get the packages from Gofer: > Gofer new > squeaksource: 'GitFS'; > package: 'FS-Git'; > package: 'Filesystem'; > load. > Cheers, > Max > > > On 09.11.2010, at 22:42, Alexandre Bergel wrote: >> Hi Max, >> >>> Um... I should be able to get a visualisation by running somthing like this right (I only installed it, didn't subclass anything)? >>> >>> MemoryProfiler >>> >> viewProfiling: [ FSDiskFilesystem current working / foo ]. >>> >>> I get a window but there's nothing in it... >> >> The piece of code you want to profile needs is executed twice, and it needs "to be time consuming" (~ 1 second). >> For example, try: >> >> MemoryProfiler >> viewProfiling: [ >> | view | >> >> view := MOViewRenderer new. >> view nodes: (1 to: 5000). >> view root applyLayout ] >> >> or >> >> MemoryProfiler >> viewProfiling: [ (1 to: 2) collect: [ :i | ConfigurationOfMetacello project currentVersion ] ] >> >> to get a nice picture >> >>> Also, is selecting the expression in the block above, right clicking and selecting "memory profile it" equivalent to sending #viewProfiling: ? >> >> yes >> >>> Reading your paper I get the idea that this should work, when looking at the instructions on the moose site it rather seems that I need to subclass MemoryProfiler to actually profile anything. >> >> No need to subclass. >> >> Cheers, >> Alexandre >> >>> >>> >>> >>> On 09.11.2010, at 15:02, Alexandre Bergel wrote: >>> >>>> Ok, but please, provide feedback. >>>> >>>> Alexandre >>>> >>>> >>>> On 9 Nov 2010, at 10:40, Max Leske wrote: >>>> >>>>> Sounds like what I just needed! I'll try it out. >>>>> >>>>> Cheers, >>>>> Max >>>>> >>>>> >>>>> On 09.11.2010, at 12:42, Alexandre Bergel wrote: >>>>> >>>>>> Hi Guido, >>>>>> >>>>>> We wrote a profiler to keep track of memory consumption. >>>>>> >>>>>> http://bergel.eu/download/papers/Berg10g-MemoryProfiling.pdf >>>>>> >>>>>> It will be presented at Smalltalks next week. >>>>>> >>>>>> Cheers, >>>>>> Alexandre >>>>>> >>>>>> On 8 Nov 2010, at 23:47, Guido Stepken wrote: >>>>>> >>>>>>> E.g.: Three different solutions for a Euler problem: >>>>>>> >>>>>>> (1 to: 999) inject: 0 into: [:sum :i | (i \\ 3 = 0) | (i \\ 5 = 0) >>>>>>> >> ifTrue: [sum + i] >>>>>>> >> ifFalse: [sum]] >>>>>>> >>>>>>> >>>>>>> (1 to: 999) detectSum: [:i | (i \\ 3 = 0) | (i \\ 5 = 0) ifTrue: [i] ifFalse: [0]] >>>>>>> >>>>>>> (1 to: 999) detectSum: [:i | i * (#(0 0 1 0 1 1 0 0 1 1 0 1 0 0 1) atWrap: i)] >>>>>>> >>>>>>> How do i measure the exact memory footprint of each solution? >>>>>>> >>>>>>> Pharo claims having reduced memory footprint at many places. Can that be proofed, e.g. with one comparable demo app, once run in old squeak, once in Pharo 1.1/1.2? >>>>>>> >>>>>>> And no, MessageTally spyOn: does not do right! >>>>>>> >>>>>>> Tnx in advance, Guido Stepken >>>>>>> >>>>>>> >>>>>> >>>>>> -- >>>>>> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: >>>>>> Alexandre Bergel http://www.bergel.eu >>>>>> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> >>>>> >>>>> >>>> >>>> -- >>>> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: >>>> Alexandre Bergel http://www.bergel.eu >>>> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. >>>> >>>> >>>> >>>> >>>> >>>> >>> >>> >> >> -- >> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: >> Alexandre Bergel http://www.bergel.eu >> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. >> >> >> >> >> >> > -- _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: Alexandre Bergel http://www.bergel.eu ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.

Nov. 13, 2010

Re: [Pharo-project] Pharo and Astronomy?

by Alexandre Bergel

Thanks all for your help on this. Cheers, Alexandre On 12 Nov 2010, at 09:08, Schwab,Wilhelm K wrote: > Alexandre, > > A possible source of a Levenberg-Marquardt optimizer: > > http://www.ics.forth.gr/~lourakis/levmar/ > > It's GPL too, so aside from _possibly_ being cleaner than GSL's implementation (there's room for improvement), it's no help. It appears that they translated the code from Minpack, and perhaps the ultimate answer is to do that "one last time" under MIT. > > Another thing I find interesting is their answer to FAQ 32. Therein find "However, it is possible to obtain a paid license for proprietary commercial use under terms different than those of GPL." Am I wrong, or can that be translated as "Pay us and you can use the code as you wish." So much for true freedom. I'm still suspicious that it is not helpful to someone wanting to release MIT wrappers of GPL code, because they are dual-licensing their code, not a wrapper. > > If nothing else, the FAQ looks interesting on topics such as computing and checking the Jacobian. > > Bill > > > > ________________________________________ > From: pharo-project-bounces(a)lists.gforge.inria.fr [pharo-project-bounces(a)lists.gforge.inria.fr] On Behalf Of Alexandre Bergel [alexandre(a)bergel.eu] > Sent: Thursday, November 04, 2010 3:14 PM > To: Pharo Development > Subject: [Pharo-project] Pharo and Astronomy? > > Hi! > > I am looking for any experience in using Pharo to develop software for Astronomy. Especially regarding interoperability with Python, using Mathematical libraries (www.nr.com, www.gnu.org/software/gsl, matplotlib.sourceforge.net/gallery.html) > > Cheers, > Alexandre > -- > _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: > Alexandre Bergel http://www.bergel.eu > ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. > > > > > > > -- _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: Alexandre Bergel http://www.bergel.eu ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.

Nov. 13, 2010