- Pharo-dev - lists.pharo.org

Re: [Pharo-project] modal windows and unit tests

by StÃ©phane Ducasse

yes this means that there is a infrastructural decision to make sure that UI can be controlled. >>> There are also some basic things that would need some adjustment as I >>> see it. I think core or basic classes and models shouldn't make any >>> assumptions about the UI or even if there is an UI at all. >> >> YESSSssssssss.... and a minicore will help /force us to do that (imagine an hudson script blocked because of a nice popup) > > Consider that this is not the UI framework that should decide to open > a dialog or not. If somebody tells morphic to open a window it should > open it otherwise you break all kinds of assumptions. > > If a user wants to support multiple UI frameworks (or the possibility > to disable certain UI actions), then he needs to take appropriate > action himself. OB for example uses notifications for all UI > interactions that are captured either by a morphic handler (by > default), a web handler (if you use the web interface), or by a test > handler (that builds a fake UI that doesn't display anything but gives > you a nice model to exercise the UI). > > Lukas > > -- > Lukas Renggli > www.lukas-renggli.ch > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

Oct. 14, 2010

[Pharo-project] Object Oriented Implementation of Numerical Methods" under the MIT license

by StÃ©phane Ducasse

I want to thanks didier for releasing the code of his book under MIT. Thanks! Begin forwarded message: > From: Didier Besset <didier(a)ieee.org> > Date: October 14, 2010 8:06:52 PM GMT+02:00 > To: StÃ©phane Ducasse <stephane.ducasse(a)inria.fr> > Subject: Disclaimer > > I hereby release the code of my book "Object Oriented Implementation of Numerical Methods" under the MIT license. > > Didier Besset

Oct. 14, 2010

Re: [Pharo-project] Stack should be reimplemented with Array

by jaayer

---- On Thu, 14 Oct 2010 10:51:31 -0700 Eliot Miranda wrote ---- >*No*. Â Stack and OrderedCollection differ *usefully*. Â Adding an item to a Stack is always O(1). Â Adding an item to an OrderedCollection is O(N) because when the array overflows a new one must be allocated and the old objects copied to the new. Â Why not apply your speedups to OrderedCollection, or is that not possible? Â If not possible, then come up with a new name and add that. Â Please /don't/ change Stack. Yes, that point was made in the prior discussion regarding Stack and its superclass in defense of having such a class at all when OrderedCollection is already there. Further, you are correct that I am suggesting we trade guaranteed O(1) complexity when pushing/popping for worst-case O(n) complexity. However, asymptotic analysis does not the whole story tell, as it ignores the underlying constants, and in the real world, when n is small enough, the constants *do* matter. As it turns out, "small enough" in this case means a Stack created with the default initial capacity of 10 grown incrementally into one big enough to hold 500,000-1,000,000 elements. I ran the following benchmark with both versions, and the array-based Stack was ~60% faster when grown big enough to contain 100,000 elements, slightly faster for 500,000 elements, and two to three times slower for a million elements: s := nil. Smalltalk garbageCollect. [s := Stack new. 100000 timesRepeat: [s push: #foo]] timeToRun. s := nil. Smalltalk garbageCollect. [s := Stack new. 500000 timesRepeat: [s push: #foo]] timeToRun. s := nil. Smalltalk garbageCollect. [s := Stack new. 1000000 timesRepeat: [s push: #foo]] timeToRun. That means that even if you unknowingly create an enormous Stack without specifying the right size, you still come out ahead for at least the first 500,000 elements. Of course, the Array-based stack never deallocates any of its memory, which means a worst case of 2 * n memory complexity (where n is the largest number of elements a given Stack ever held). But as I said in my initial post, most Stacks are small, and when they aren't, you often have enough information to choose an appropriate size. If you really need constant time and memory complexity, which I argue is the exception, why not just use LinkedList directly with #addLast: and #removeLast to simulate a Stack as people did and still do with OrderedCollection? The rueful requirement that LinkList elements inherit from Link is gone. Stack, if it is to exist at all, should support *fast* push and pop operations. The current implementation does not. This one does, and should replace it.

Oct. 14, 2010

[Pharo-project] Why atomic stuff rules :)

by Igor Stasenko

In the attachment an implementation of SharedQueue3, which using atomic operation. Of course , currently it is based on assumption that some stuff (like two assignments in a row) can't be interrupted. The FIFO implementation using ASWP and CAS: push: - no spins, 1 ASWP pop: - 2 spins, 1 ASWP , 1 CAS Here a simple bench i made, which pushing 100000 elements into queue using 10 different processes, and then looks how many are popped via 3 feeder processes during that time: Squeak VM. SharedQueue : SharedQueue3Test new benchManyPushersAndFeeders: SharedQueue 'Added 100000 items, processed 30000 queue items during 1928 milliseconds' SharedQueue2 : SharedQueue3Test new benchManyPushersAndFeeders: SharedQueue2 'Added 100000 items, processed 30034 queue items during 1074 milliseconds' SharedQueue3 (mine) : SharedQueue3Test new benchManyPushersAndFeeders: SharedQueue3 'Added 100000 items, processed 70229 queue items during 118 milliseconds' Cog: SharedQueue3Test new benchManyPushersAndFeeders: SharedQueue 'Added 100000 items, processed 29999 queue items during 1955 milliseconds' SharedQueue3Test new benchManyPushersAndFeeders: SharedQueue2 'Added 100000 items, processed 29997 queue items during 197 milliseconds' SharedQueue3Test new benchManyPushersAndFeeders: SharedQueue3 'Added 100000 items, processed 100000 queue items during 26 milliseconds' if anyone has better ideas how to bench throughput, send me the code. -- Best regards, Igor Stasenko AKA sig.

Oct. 14, 2010

[Pharo-project] OOP best practices

by Fernando olivero

Hi Hernan, just wanted to get your opinion on the following method, that attempts to adhere to your "complete objects" pattern. ( ESUG 2010 TALK). PerformedExperiment>>experiment: anExperiment start: aTime stop: aStopTime participant: aParticipant tasks: aCollectionOfTasks | experiment | experiment := self new. experiment experiment: anExperiment start: aTime stop: aStopTime participant: aParticipant tasks: aCollectionOfTasks. ^ experiment In your experience, how does the pattern cope with large keyword selectors. Maybe this case is not that evident, although we should try to avoid having more than 4 instance variables anyway, i would like to get your opinion on this problem (?). Thanks, Fernando

Oct. 14, 2010

Re: [Pharo-project] Stack should be reimplemented with Array

by jaayer

The original contains an error; it send min: where it should max:. This one contains the correction.

Oct. 14, 2010

Re: [Pharo-project] Stack should be reimplemented with Array

by Eliot Miranda

2010/10/14 jaayer <jaayer(a)zoho.com> > There was a post on this list back in August complaining about Stack > inheriting from LinkedList. There was some discussion, but apparently no > resolution, as Stack still inherits from LinkedList in the Pharo 1.2 Core > Image I just downloaded. Rather than reimplementing it to forward to a > contained LinkedList, I think it should use an Array internally like > OrderedCollection does. An Array-based implementation of Stack would be > faster than one based on LinkedList due to the basic stack operations (push, > pop and top) being able to rely more on primitives. This assumes that you > know roughly how a big a stack you need; if you don't, reallocating the > array and copying its contents over and over again will cost you, but this > is an exceptional case;-- most stacks are small and when they aren't, the > programmer usually has enough information to select a reasonable stack size. > *No*. Stack and OrderedCollection differ *usefully*. Adding an item to a Stack is always O(1). Adding an item to an OrderedCollection is O(N) because when the array overflows a new one must be allocated and the old objects copied to the new. Why not apply your speedups to OrderedCollection, or is that not possible? If not possible, then come up with a new name and add that. Please /don't/ change Stack. > > Enclosed is a fileout of an Array-based Stack implementation. It trounces > the LinkedList-based implementation easily, but more interestingly, it > performs better than an OrderedCollection when used like a stack: > > Pushing is roughly ~30% faster: > r1 :=[100000 timesRepeat: [ > s := Stack new. > 10 timesRepeat: [s push: #foo]]] timeToRun. > r2 := [100000 timesRepeat: [ > oc := OrderedCollection new. > 10 timesRepeat: [oc addLast: #foo]]] timeToRun. > 100 - ((r1 / r2) asFloat * 100). > > as is pushing + popping: > r3 := [100000 timesRepeat: [ > s := Stack new. > 10 timesRepeat: [s push: #foo]. > 10 timesRepeat: [s pop]]] timeToRun. > r4 := [100000 timesRepeat: [ > oc := OrderedCollection new. > 10 timesRepeat: [oc addLast: #foo]. > 10 timesRepeat: [oc removeLast]]] timeToRun. > 100 - ((r3 / r4) asFloat * 100). > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project >

Oct. 14, 2010

Re: [Pharo-project] Bug when changing superclass in latest OB?

by Simon Denier

On 14 oct. 2010, at 19:37, Lukas Renggli wrote: > Hi Simon, > > It works for me too, the change can even be undone and redone. Also > this should be tested as part of the unit tests. > > Maybe try to load the code again and recompile all packages: > > Gofer new > squeaksource: 'rb'; > package: 'AST-Core'; > package: 'Refactoring-Core'; > package: 'Refactoring-Changes'; > package: 'Refactoring-Critics'; > package: 'Refactoring-Environment'; > load; > recompile. > > Gofer new > renggli: 'omnibrowser'; > package: 'OmniBrowser'; > package: 'OB-Standard'; > package: 'OB-Morphic'; > package: 'OB-Refactory'; > load; > recompile Indeed it works as expected when I reload+recompile. Thanks Lukas. > > > On 14 October 2010 19:23, Alexandre Bergel <alexandre(a)bergel.eu> wrote: >> hi Simon, >> >> I cannot reproduce. It works as expected for me. >> >> Alexandre >> >> >> On 14 Oct 2010, at 12:43, Simon Denier wrote: >> >>> Hi >>> >>> Downloading the latest OB today with: >>> >>> Gofer new >>> squeaksource: 'rb'; >>> package: 'AST-Core'; >>> package: 'Refactoring-Core'; >>> package: 'Refactoring-Changes'; >>> package: 'Refactoring-Critics'; >>> package: 'Refactoring-Environment'; >>> load. >>> >>> Gofer new >>> renggli: 'omnibrowser'; >>> package: 'OmniBrowser'; >>> package: 'OB-Standard'; >>> package: 'OB-Morphic'; >>> package: 'OB-Refactory'; >>> load. >>> >>> >>> Then I change the superclass of the template of an existing class, then accept: >>> >>> Object subclass: #FileContentsBrowser >>> instanceVariableNames: 'packages infoString' >>> classVariableNames: '' >>> poolDictionaries: '' >>> category: 'Tools-File Contents Browser' >>> >>> >>> The superclass is not changed, instead OB changes its focus to the superclass. >>> >>> Can anyone confirm this bug? >>> >>> >>> -- >>> Simon >>> >>> >>> >>> >>> _______________________________________________ >>> Pharo-project mailing list >>> Pharo-project(a)lists.gforge.inria.fr >>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project >> >> -- >> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: >> Alexandre Bergel http://www.bergel.eu >> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. >> >> >> >> >> >> >> _______________________________________________ >> Pharo-project mailing list >> Pharo-project(a)lists.gforge.inria.fr >> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project >> > > > > -- > Lukas Renggli > www.lukas-renggli.ch > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project -- Simon

Oct. 14, 2010

[Pharo-project] Stack should be reimplemented with Array

by jaayer

There was a post on this list back in August complaining about Stack inheriting from LinkedList. There was some discussion, but apparently no resolution, as Stack still inherits from LinkedList in the Pharo 1.2 Core Image I just downloaded. Rather than reimplementing it to forward to a contained LinkedList, I think it should use an Array internally like OrderedCollection does. An Array-based implementation of Stack would be faster than one based on LinkedList due to the basic stack operations (push, pop and top) being able to rely more on primitives. This assumes that you know roughly how a big a stack you need; if you don't, reallocating the array and copying its contents over and over again will cost you, but this is an exceptional case;-- most stacks are small and when they aren't, the programmer usually has enough information to select a reasonable stack size. Enclosed is a fileout of an Array-based Stack implementation. It trounces the LinkedList-based implementation easily, but more interestingly, it performs better than an OrderedCollection when used like a stack: Pushing is roughly ~30% faster: r1 :=[100000 timesRepeat: [ s := Stack new. 10 timesRepeat: [s push: #foo]]] timeToRun. r2 := [100000 timesRepeat: [ oc := OrderedCollection new. 10 timesRepeat: [oc addLast: #foo]]] timeToRun. 100 - ((r1 / r2) asFloat * 100). as is pushing + popping: r3 := [100000 timesRepeat: [ s := Stack new. 10 timesRepeat: [s push: #foo]. 10 timesRepeat: [s pop]]] timeToRun. r4 := [100000 timesRepeat: [ oc := OrderedCollection new. 10 timesRepeat: [oc addLast: #foo]. 10 timesRepeat: [oc removeLast]]] timeToRun. 100 - ((r3 / r4) asFloat * 100).

Oct. 14, 2010

Re: [Pharo-project] Bug when changing superclass in latest OB?

by Lukas Renggli

Hi Simon, It works for me too, the change can even be undone and redone. Also this should be tested as part of the unit tests. Maybe try to load the code again and recompile all packages: Gofer new squeaksource: 'rb'; package: 'AST-Core'; package: 'Refactoring-Core'; package: 'Refactoring-Changes'; package: 'Refactoring-Critics'; package: 'Refactoring-Environment'; load; recompile. Gofer new renggli: 'omnibrowser'; package: 'OmniBrowser'; package: 'OB-Standard'; package: 'OB-Morphic'; package: 'OB-Refactory'; load; recompile On 14 October 2010 19:23, Alexandre Bergel <alexandre(a)bergel.eu> wrote: > hi Simon, > > I cannot reproduce. It works as expected for me. > > Alexandre > > > On 14 Oct 2010, at 12:43, Simon Denier wrote: > >> Hi >> >> Downloading the latest OB today with: >> >> Gofer new >> Â Â Â squeaksource: 'rb'; >> Â Â Â package: 'AST-Core'; >> Â Â Â package: 'Refactoring-Core'; >> Â Â Â package: 'Refactoring-Changes'; >> Â Â Â package: 'Refactoring-Critics'; >> Â Â Â package: 'Refactoring-Environment'; >> Â Â Â load. >> >> Gofer new >> Â Â Â renggli: 'omnibrowser'; >> Â Â Â package: 'OmniBrowser'; >> Â Â Â package: 'OB-Standard'; >> Â Â Â package: 'OB-Morphic'; >> Â Â Â package: 'OB-Refactory'; >> Â Â Â load. >> >> >> Then I change the superclass of the template of an existing class, then accept: >> >> Object subclass: #FileContentsBrowser >> Â Â Â instanceVariableNames: 'packages infoString' >> Â Â Â classVariableNames: '' >> Â Â Â poolDictionaries: '' >> Â Â Â category: 'Tools-File Contents Browser' >> >> >> The superclass is not changed, instead OB changes its focus to the superclass. >> >> Can anyone confirm this bug? >> >> >> -- >> Simon >> >> >> >> >> _______________________________________________ >> Pharo-project mailing list >> Pharo-project(a)lists.gforge.inria.fr >> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project > > -- > _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: > Alexandre Bergel Â http://www.bergel.eu > ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. > > > > > > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project > -- Lukas Renggli www.lukas-renggli.ch

Oct. 14, 2010