March 2012 - Pharo-dev - lists.pharo.org

Re: [Pharo-project] [Moose-dev] numerical method in smalltalk

by Serge Stinckwich

On Tue, Mar 20, 2012 at 12:45 PM, Tudor Girba <tudor(a)tudorgirba.com> wrote: > Hi Serge, > > There definitely is interest. If you volunteer for mentoring it would be great. Yes sure i can do it. I can even maybe propose some local students. Do you want to co-mentor ? Regards, -- Serge Stinckwich UMI UMMISCO 209 (IRD/UPMC), Hanoi, Vietnam Every DSL ends up being Smalltalk http://doesnotunderstand.org/

March 20, 2012

Re: [Pharo-project] [Moose-dev] numerical method in smalltalk

by Tudor Girba

Hi Serge, There definitely is interest. If you volunteer for mentoring it would be great. Cheers, Doru On 20 Mar 2012, at 04:35, Serge Stinckwich wrote: > Hi all, > > we talked recently on the moose mailing-list about having a more > robust library for doing mathematical stuff in Smalltalk > like SciPython or SciRuby (see below). If there is enough interested, > i'm wondering if we could propose a Google Summer of Code project > about that. I could write a draft for the project. > > Regards, > > On Mon, Mar 5, 2012 at 10:05 AM, Serge Stinckwich > <serge.stinckwich(a)gmail.com> wrote: >> On Sat, Mar 3, 2012 at 4:45 AM, StÃ©phane Ducasse >> <Stephane.Ducasse(a)inria.fr> wrote: >>> could be interesting to see if we need to some numerical analysis in moose and pharo. >>> >>> I will create a configuration >>> >>> >>> MCSqueaksourceRepository >>> location: 'http://squeaksource.com/DHBNumerical' >>> user: '' >>> password: '' >> >> Thank you. I'm definitively interested by having more mathematical >> stuff available in Pharo. >> There is Scipy [1] since a long time for Python and more recently Ruby >> community do the same with SciRuby [2]. >> >> Some time ago, i wrote several random number generators [3] that might >> be integrated in a numeral analysis packages. >> >> At the moment, i'm mostly interested by Runge-Kutta methods [4] in >> order to solve some ordinary differential equations. >> I already find some code on the web like this one: >> http://live.exept.de/doc/online/english/programming/goody_stmath.html >> >> What is really important is to be able to test the result of such algorithms. >> >> [1] https://en.wikipedia.org/wiki/SciPy >> [2] http://sciruby.com/ >> [3] http://www.squeaksource.com/Random.html >> [4] https://en.wikipedia.org/wiki/Runge%E2%80%93Kutta_methods >> >> Regards, >> -- >> Serge Stinckwich >> UMI UMMISCO 209 (IRD/UPMC), Hanoi, Vietnam >> Every DSL ends up being Smalltalk >> http://doesnotunderstand.org/ > > > > -- > Serge Stinckwich > UMI UMMISCO 209 (IRD/UPMC), Hanoi, Vietnam > Every DSL ends up being Smalltalk > http://doesnotunderstand.org/ > -- www.tudorgirba.com "Not knowing how to do something is not an argument for how it cannot be done."

March 20, 2012

Re: [Pharo-project] Reliable method change detection

by Martin Dias

Finally I tried with a bytecodes checksum to detect changes. I chose a 16-bit variant of Adler-32 (http://en.wikipedia.org/wiki/Adler-32) Was easy and worked fine. ByteArray >> adler16from: start to: end "Answer a checksum over the values of the specified interval, calculated using Adler-16 algorithm." | a b | a := 1. b := 0. start to: end do: [ :index | a := (a + (self at: index)) \\ 251. b := (b + a) \\ 251. ]. ^(b bitShift: 8) bitOr: a CompiledMethod >> bytecodesHash "Answer a 16-bit checksum of the bytecodes." ^ self adler16from: self initialPC to: self endPC cheers, MartÃn On Sat, Mar 17, 2012 at 2:17 AM, Martin Dias <tinchodias(a)gmail.com> wrote: > Hi > > In Fuel (bleeding edge), a compiled method can be serialized as a global. > Basically, we store the class name and selector, so loading it is > straightforward... this can be fine if we suppose there were not any change > in the system between save and load time. This can be a huge supposition, > but maybe it's acceptable in some uses, for example to "log" exceptions. > > Well, actually we can discuss if this is a good approach, but anyway I am > interested on how to detect a BIG change, for example the one I show in the > following code (which produces a vm crash): > > | byteArray aClass aClosure materializedClosure | > > aClass := Object subclass: #A. > aClass compileSilently: 'methodWithClosure ^ [ 42 ] '. > > aClosure := aClass new perform: #methodWithClosure. > > byteArray := FLSerializer serializeToByteArray: aClosure. > > aClass compileSilently: 'methodWithClosure ^ 42'. > > materializedClosure := FLMaterializer materializeFromByteArray: byteArray. > materializedClosure value. > > > What can we do to detect this change? > - generate a checksum with bytecodes > - just compare the size > - something else? > > Thanks, > MartÃn >

March 20, 2012

Re: [Pharo-project] [Moose-dev] numerical method in smalltalk

by Serge Stinckwich

Hi all, we talked recently on the moose mailing-list about having a more robust library for doing mathematical stuff in Smalltalk like SciPython or SciRuby (see below). If there is enough interested, i'm wondering if we could propose a Google Summer of Code project about that. I could write a draft for the project. Regards, On Mon, Mar 5, 2012 at 10:05 AM, Serge Stinckwich <serge.stinckwich(a)gmail.com> wrote: > On Sat, Mar 3, 2012 at 4:45 AM, StÃ©phane Ducasse > <Stephane.Ducasse(a)inria.fr> wrote: >> could be interesting to see if we need to some numerical analysis in moose and pharo. >> >> I will create a configuration >> >> >> MCSqueaksourceRepository >> Â Â Â Â location: 'http://squeaksource.com/DHBNumerical' >> Â Â Â Â user: '' >> Â Â Â Â password: '' > > Thank you. I'm definitively interested by having more mathematical > stuff available in Pharo. > There is Scipy [1] since a long time for Python and more recently Ruby > community do the same with SciRuby [2]. > > Some time ago, i wrote several random number generators [3] that might > be integrated in a numeral analysis packages. > > At the moment, i'm mostly interested by Runge-Kutta methods [4] in > order to solve some ordinary differential equations. > I already find some code on the web like this one: > http://live.exept.de/doc/online/english/programming/goody_stmath.html > > What is really important is to be able to test the result of such algorithms. > > [1] https://en.wikipedia.org/wiki/SciPy > [2] http://sciruby.com/ > [3] http://www.squeaksource.com/Random.html > [4] https://en.wikipedia.org/wiki/Runge%E2%80%93Kutta_methods > > Regards, > -- > Serge Stinckwich > UMI UMMISCO 209 (IRD/UPMC), Hanoi, Vietnam > Every DSL ends up being Smalltalk > http://doesnotunderstand.org/ -- Serge Stinckwich UMI UMMISCO 209 (IRD/UPMC), Hanoi, Vietnam Every DSL ends up being Smalltalk http://doesnotunderstand.org/

March 20, 2012

Re: [Pharo-project] MVP - presenters as factories; some models; CODE!!!

by Schwab,Wilhelm K

I was playing with a workbench that I built for a long-standing project with LOTS of data, and realized that I was leaving a trail of realized (read chewing up a lot of memory) objects as I tinkered. That underscored the need for #selectionChanging: on trees and lists. The attached appears to successfully add same, though it has not had much testing. Try the cartoon example. You should now be able to edit the fields and see the results of the change the next time you visit the character. For example, Bullwinkle's IQ is listed as low. You might want to change that to VERY LOW :) Try it. Bill ________________________________ From: pharo-project-bounces(a)lists.gforge.inria.fr [pharo-project-bounces(a)lists.gforge.inria.fr] on behalf of Schwab,Wilhelm K [bschwab(a)anest.ufl.edu] Sent: Monday, March 19, 2012 9:44 PM To: pharo-project(a)lists.gforge.inria.fr Subject: [Pharo-project] MVP - presenters as factories; some models; CODE!!! Hello all, I am a strong believer in "say it once," and I find Polymorph's lists, and especially trees, very cumbersome to use. One is led to create a slot to hold selections and accessors. That work should be done by by a reusable presenter with some help from ListModel and TreeModel. Attached is the current state (of confusion) of the framework. The factory concept (presenters create morph-model pairs, the views are all morphs) is new, in some sense a concession to reality. I would prefer other layout means (see the various "obsolete" parts of the package), but have not been able to make them work. Using polymorph rows and columns, I can get close enough to what I need, at least with a suitable user base. The GUI is nothing special, but the way it is built is, I think, worthy of some attention. Try CartoonPersonalityBrowser new asMorph openInWorld. for a few laughs. For those not familiar with it, the Rocky and Bullwinkle Show was *not* for kids. Kids will enjoy it, but it was in fact beautiful cold-war satire. The road runner and coyote are more likely to be familiar to you, and are just good (great) slapstick with a few comments on life (everything works fine until the roadrunner appears...). The list and tree scream for a #selectionChanging: event, which I had not appreciated until I created the above example. In short, you can edit the fields, but they are not saved. Punching the presenters onto the tree elements would be a little tricky, but one should be able to know that the selection is changing, react to it (saving current contents) and then even veto the change. More work... Please comment freely. Bill

March 20, 2012

[Pharo-project] MVP - presenters as factories; some models; CODE!!!

by Schwab,Wilhelm K

Hello all, I am a strong believer in "say it once," and I find Polymorph's lists, and especially trees, very cumbersome to use. One is led to create a slot to hold selections and accessors. That work should be done by by a reusable presenter with some help from ListModel and TreeModel. Attached is the current state (of confusion) of the framework. The factory concept (presenters create morph-model pairs, the views are all morphs) is new, in some sense a concession to reality. I would prefer other layout means (see the various "obsolete" parts of the package), but have not been able to make them work. Using polymorph rows and columns, I can get close enough to what I need, at least with a suitable user base. The GUI is nothing special, but the way it is built is, I think, worthy of some attention. Try CartoonPersonalityBrowser new asMorph openInWorld. for a few laughs. For those not familiar with it, the Rocky and Bullwinkle Show was *not* for kids. Kids will enjoy it, but it was in fact beautiful cold-war satire. The road runner and coyote are more likely to be familiar to you, and are just good (great) slapstick with a few comments on life (everything works fine until the roadrunner appears...). The list and tree scream for a #selectionChanging: event, which I had not appreciated until I created the above example. In short, you can edit the fields, but they are not saved. Punching the presenters onto the tree elements would be a little tricky, but one should be able to know that the selection is changing, react to it (saving current contents) and then even veto the change. More work... Please comment freely. Bill

March 20, 2012

Re: [Pharo-project] Object>>deprecated:on:in: doesn't respect state of Deprecation class>>showWarning:

by Martin Sandiford

OK, done. http://code.google.com/p/pharo/issues/detail?id=5521 On Tue, Mar 20, 2012 at 8:25 AM, Fernando Olivero <fernando.olivero(a)usi.ch> wrote: > Hi Martin, > no objections! > please create the issue, and attach the fix. > Fernando

March 19, 2012

Re: [Pharo-project] non-interactive compiling with feedback

by Eliot Miranda

On Mon, Mar 19, 2012 at 1:11 PM, Milan Mimica <milan.mimica(a)gmail.com>wrote: > On 19 March 2012 21:04, Eliot Miranda <eliot.miranda(a)gmail.com> wrote: > >> >> a) compiling source containing an undeclared reference merits only a >> warning >> >> IMO, it's not an error, and a) is a good option. Note that if you did >> make it an error in all circumstances you'd badly break the system. >> > > > Hi Eliot, > > a) works for me. I just need a string to display to the user, on compile > time. So how do I get a warning? > Look at UndeclaredVariableWarning. The Squeak code isn't well favtored. defaultAction needs to be decomposed further into a messageText implementation. But basically I'd want the UndeclaredVariableWarning's messageText to provide the warning string. So I could do e.g. [Compiler evaluate: aString] on: UndeclaredVariableWarning do: [:ex| Transcript cr; show: ex messageText] > > > -- > Milan Mimica > http://sparklet.sf.net > -- best, Eliot

March 19, 2012

Re: [Pharo-project] "Browse class" : Nautilus class DNU #newOnClass:

by Benjamin

Thanks for your report :) Ben On Mar 17, 2012, at 6:21 PM, Alexis Parseghian wrote: > Fix by implementing it and have it call Nautilus class>>openOnClass: > > http://code.google.com/p/pharo/issues/detail?id=5505 >

March 19, 2012

Re: [Pharo-project] Object>>deprecated:on:in: doesn't respect state of Deprecation class>>showWarning:

by Fernando Olivero

Hi Martin, no objections! please create the issue, and attach the fix. Fernando On Mon, Mar 19, 2012 at 10:47 PM, Martin Sandiford <ms(a)mcdev.com.au> wrote: > Hi all, > > Kind of new to Pharo, so not sure if this is intended behaviour or > not, but "Object>>deprecated:on:in:" doesn't respect the current state > of the "Deprecation class>>showWarning" setting. > > This was unexpected for me --- in a headless setting the warning > dialog causes the VM to exit. > > To reproduce (in 1.4 14399): > > Deprecation showWarning: false. > true and: [ true ] and: [ true ]. "Boolean>>and:and: sends deprecated:on:in:" > > Fix seems straightforward: wrap the current contents of > "deprecated:on:in:" in "Deprecation showWarning ifTrue: [...]". > > If there are no objections, I'll raise a bug in the tracker. > > Cheers, > Martin >

March 19, 2012