November 2015 - Pharo-users

Re: [Pharo-users] Enable/Disable GTSpotter

by stepharo

Le 18/11/15 13:56, Esteban Lorenzano a Ã©crit : >> On 18 Nov 2015, at 09:50, Tudor Girba <tudor(a)tudorgirba.com> wrote: >> >> Hi, >> >> Ok, for that problem we already have a solution provided by the Keymapping engine: >> >> (KMRepository default globalCategories detect: [ :each | >> each isKindOf: GTSpotterGlobalShortcut ]) >> allEntries keymaps do: #disable > thatâs a workaround, not a solution :) > (Iâm ok if thatâs the way to do it in Pharo4â¦ I will still add a setting in Pharo5) And a unload method. > > cheers, > Esteban > >> Cheers, >> Doru >> >> >>> On Nov 18, 2015, at 1:38 PM, Ferlicot D. Cyril <cyril.ferlicot(a)gmail.com> wrote: >>> >>> Le 18/11/2015 13:25, Tudor Girba a Ã©crit : >>>> Do you mean that you want to have a different search solution when you press Shift+Enter or when you select it from the World menu? >>>> >>>> At the moment, this is not possible. >>>> >>>> May I ask where is the problem coming from? Perhaps there are things we can do to solve your problem. >>>> >>> The problem is that I want a desktop application. >>> >>> With the spotter we can access to the code, so I just want to disable it >>> when we are not in development mode. >>> >>> We can add a setting to Pharo 5 but in Pharo 4 there is no easy way to >>> disable it. >>> >>> I think we should also add an easy way to create a deployment image >>> directly inside Pharo. >>> >>>> Cheers, >>>> Doru >>>> >>>> -- >>>> www.tudorgirba.com >>>> >>>> "Not knowing how to do something is not an argument for how it cannot be done." >>>> >>>> >>> >>> -- >>> Cyril Ferlicot (cyril.ferlicot(a)synectique.eu) >>> >>> http://www.synectique.eu >>> >>> 165 Avenue Bretagne >>> Lille 59000 France >>> >> -- >> www.tudorgirba.com >> >> "Reasonable is what we are accustomed with." >> >> > >

Nov. 19, 2015

Re: [Pharo-users] BitField

by stepharo

Have a look at the chapter on number in deep into pharo. You get some illustrations of bit operations for numbers bitShift: bitAnd: ... Stef Le 16/11/15 21:16, abdelghani ALIDRA a Ã©crit : > Hi everybody, > > I was discussing with Thibault about the performance of some methods > defined on sets ( actually quite slow for big collections, mainly > union: and intersection: ) and we were wondering if there is any > implementation of some BitField-like class to speed up these > operations? May be in an external project? > > Abdelghani

Nov. 19, 2015

Re: [Pharo-users] [Pharo-dev] Possible dates for PharoDays 2016

by Alexandre Bergel

No idea if I can attend, but beginning of March is better for me. Alexandre -- _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: Alexandre Bergel http://www.bergel.eu ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. > On Nov 19, 2015, at 3:57 PM, stepharo <stepharo(a)free.fr> wrote: > > Hi > > Okay dates would be 3 & 4 /3 or 31/3 & 1/4 > And the location would be at Namur in Belgium > > Could you tell us if there is a problem for one of these dates? > > Stef >

Nov. 19, 2015

Re: [Pharo-users] BitField

by Marcus Denker

Hello, You maybe can use Integers as bitfield: bitField := 0 bitAt: 200 put: 1. executing bitField bitAt: 200 will return 1. The only thing to take care is that it # bitAt:put does not change the object itself, but instead returns a new one. Marcus > On 16 Nov 2015, at 17:16, abdelghani ALIDRA <alidrandco(a)yahoo.fr> wrote: > > Hi everybody, > > I was discussing with Thibault about the performance of some methods defined on sets ( actually quite slow for big collections, mainly union: and intersection: ) and we were wondering if there is any implementation of some BitField-like class to speed up these operations? May be in an external project? > > Abdelghani

Nov. 19, 2015

Re: [Pharo-users] Array : very weird behavior

by Marcus Denker

Hello, You are changing the âliteral arrayâ itself. The #(0 0 0 0) creates an array as part of the compiledMethod and you are changing this. (Yes, many people think that all literals should be immutableâ¦). If you add a #copy it will work. bitArray1 := #(0 0 0 0) copy or create the array at runtime with {0.0.0.0} As for bit array (bit field), I wanted to answer your original question which I will do laterâ¦ Marcus > On 19 Nov 2015, at 16:12, abdelghani ALIDRA <alidrandco(a)yahoo.fr> wrote: > > Hi everybody, > > I am experiencing a very weird behavior with Arrays : > > I writed a test case as follows ; > > testChange > | bitArray1 | > > > bitArray1 := #(0 0 0 0). > > (1 to: bitArray1 size) > do: [ :index ||temp| > temp := (bitArray1 at: index) + 1. > bitArray1 at: index put: (temp). > ]. > self assert: bitArray1 = #(1 1 1 1) > > > The thing is that the test succedes the first time but if I run it again it fails. > Moreover, if I change the testChange arbitrarily (just add a space and save) the test succeeds the first time then it fails.....etc. > > Actually bitArray1 := #(0 0 0 0) changes the value of bitArray1 in the first run but then bitArray1 does not change any more in the next runs. > > The problem does not occur if I replace bitArray1 := #(0 0 0 0). with bitArray1 := Array new:4 withAll: 0 . > > Any idea where the problem can be? > Should I open a bug entry? where? > > Abdelghani >

Nov. 19, 2015

Re: [Pharo-users] Stuck at lookup implementation in "Building ObjVlisp" tutorial

by stepharo

I checked testNilWhenErrorInLookup "(self selector: #testNilWhenErrorInLookup) run" self assert: (pointClass lookup: #zork) isNil. "The method zork is NOT implement on pointClass" testRaisesErrorSendWhenErrorInLookup "(self selector: #testRaisesErrorSendWhenErrorInLookup) run" self should: [ pointClass send: #zork withArguments: { aPoint } ] raise: Error. "Open a Transcript to see the message trace" Le 19/11/15 13:50, Edouard Delbar a Ã©crit : > Hello everyone, > > I'm new on this mailing list, this is my first mail :-) > > I'm a student from University of Lille (France). I'm studying this > tutorial (I'm using Pharo 4 but it seems not to be a problem): > http://sdmeta.gforge.inria.fr/Teaching/ObjVLisp/ObjV.pillar.pdf. I'm > working with the "ObjVLispSkeleton-StephaneDucasse.1" image loaded > from this repository: > > MCSmalltalkhubRepository > owner: 'StephaneDucasse' > project: 'ObjVLispSkeleton' > user: '' > password: '' > > > I'm stuck at "Method Lookup" p11. First, it seems that unit tests > "testNilWhenErrorInLookup" and "testRaisesErrorSendWhenErrorInLookup" > are not provided, thus I'm not sure that I'm working with the right > image (everything was fine until p11 though). Then, the "lookup: > selector" is not provided neither. While there is another method named > "lookup: selector from: anObjObject", I can't figure out what I'm > supposed to do with the second parameter. Ok I checked in the other method is old the new one should be lookup: selector "look for the method named <selector> starting in the receiver. The lookup is done for a message sent to <anObjObject>. self is an objClass" > Here is what I would do: > > lookup: selector > "look for the method named <selector> starting in the receiver. > The lookup is done for a message sent to <anObjObject>. self is an > objClass" > ^ self objMethodDict > at: selector > ifAbsent: [ self objSuperclassId ifNotNil: [ self objSuperclass > lookup: selector ] ] > > > Could someone tell me if this implementation is correct? If it's > correct, what do you think the second parameter is for? > > Moreover, do you think I'm working with the right image? > > Thanks. > > - Edouard

Nov. 19, 2015

Re: [Pharo-users] Stuck at lookup implementation in "Building ObjVlisp" tutorial

by stepharo

Hi edouard I will check as soon as I get some time. > I'm new on this mailing list, this is my first mail :-) welcome > I'm a student from University of Lille (France). I'm studying this > tutorial (I'm using Pharo 4 but it seems not to be a problem): > http://sdmeta.gforge.inria.fr/Teaching/ObjVLisp/ObjV.pillar.pdf. I'm > working with the "ObjVLispSkeleton-StephaneDucasse.1" image loaded > from this repository: > > MCSmalltalkhubRepository > owner: 'StephaneDucasse' > project: 'ObjVLispSkeleton' > user: '' > password: '' > > > I'm stuck at "Method Lookup" p11. First, it seems that unit tests > "testNilWhenErrorInLookup" and "testRaisesErrorSendWhenErrorInLookup" > are not provided, Mark if you read this mail. Did you get problem when you did it? > thus I'm not sure that I'm working with the right image (everything > was fine until p11 though). Then, the "lookup: selector" is not > provided neither. Not providing lookup: is ok since you have to define it. I will look for the tests in my version. > While there is another method named "lookup: selector from: > anObjObject", I can't figure out what I'm supposed to do with the > second parameter. What you should see is that the second argument is is provided here because when you raise an error, you should send a messageNotUnderstood and this message should be sent to the original object that received the message and not the class where the method was found. So lookup will use the receiver to walk through the inheritance chain and the from: argument in case of error > Here is what I would do: > > lookup: selector > "look for the method named <selector> starting in the receiver. > The lookup is done for a message sent to <anObjObject>. self is an > objClass" > ^ self objMethodDict > at: selector > ifAbsent: [ self objSuperclassId ifNotNil: [ self objSuperclass > lookup: selector ] ] > what do you return when you do not find the class > > Could someone tell me if this implementation is correct? If it's > correct, what do you think the second parameter is for? > > Moreover, do you think I'm working with the right image? > > Thanks. > > - Edouard

Nov. 19, 2015

Re: [Pharo-users] More fun with Roassal: 1st iteration of a Roassal Mindmapping tool

by Alexandre Bergel

Looks good!!!! You are @embee on Slack right ? :-) Cheers, Alexandre -- _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: Alexandre Bergel http://www.bergel.eu ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. > On Nov 19, 2015, at 4:03 PM, Martin BÃ¤hr <mbaehr(a)email.archlab.tuwien.ac.at> wrote: > > Excerpts from stepharo's message of 2015-08-22 09:55:04 +0200: >> would be fun to see if you can reuse the algo of roassal. > > i have taken a stab at that: > > =========================================================================================== > view := RTView new. > | activenode prevnode | > > view > when: TRMouseClick > do: > [ :ev | > | label node | > "node := shape element." > label := RTEditableLabel new > height: 30; "i have a very high resolution and like large fonts" > text: ''; > element. > node := RTBox new > color: Color white; > borderWidth: 2; > borderColor: Color black; > element. > node @ RTDraggable. > > "select a node to branch of new nodes from it" > node when: TRMouseClick do: [ :click | activenode := click element ]. > > "if the label is empty: '', then its size is 0@0 and is no > longer clickable to edit, hence we make the node clickable instead" > node when: TRMouseDoubleClick do: [ label shape showTextMorphFor: label ]. > node when: TRMouseLeave do: [ label shape hideTextMorphFor: label ]. > view > add: node; > add: label. > RTNest new > stretchable; > on: node nest: {label}. > > "when the label changes its size, the box doesn't update its > size unless we add this workaround" > label trachelShape addCallback: (TRExtentCallback new block: [ :aShape :ratio | node trachelShape extent: aShape extent + (20 @ 0) ]). > > node translateTo: ev position. > > "label shape showTextMorphFor: label." "this opens the node in edit mode, > but somehow messes things up" > > (activenode notNil and: node ~= activenode) > ifTrue: [ prevnode := activenode ]. > prevnode > ifNotNil: [ edge := RTEdge from: prevnode to: node. > edge > + > (RTBezierLine horizontal > withContinuousRectangleAttachPoint; > yourself). > view add: edge ]. > prevnode := node. > activenode := node. > view signalUpdate ]. > view. > =========================================================================================== > > the behavior is a bit different than the usual mindmapping tools as it does not > enforce an automatic layout but you are free to place nodes whereever you like > by just clicking in the location and moving the node later. > > that behavior is inspired by a mindmapping tool called labyrint: > https://github.com/labyrinth-team/labyrinth > https://www.youtube.com/watch?v=Pv3dTovGc0E > > > thanks to alexandre bergel and others who helped me to figure out some > problems such as a workaround for RTEditableLabel not working with > RTCompositeShape: > > shape := RTCompositeShape new. > shape allOfSameSizeWithPadding. > shape add: (RTBox new color: Color white; borderWidth: 2; borderColor: Color black). > shape add: (RTEditableLabel new height: 30). > > the remaining problems are cosmetic. for example i can't seem to get the nodes > to open up in edit mode. doing that with the commented line in the code above, > wrecks havoc on the node. > i also would like RTBezierLine to be horizontal or vertical depending on how > the two nodes are positioned to each other. and it should always start at the > middle of the rectangles edge at a 90degree angle. > > lastly, i'd like to have rounded corners on the RTBox. > > if/when i develop this further, i plan to add the ability to select multiple > nodes and add edges between them. be able to move selected nodes as a group, > and align selected (or all) nodes according to one of the layouts provided by > roassal (tree, circle, etc), things like export to svg or png would also be nice. > > and of course, package it up... > > greetings, martin. > > -- > eKita - the online platform for your entire academic life > -- > chief engineer eKita.co > pike programmer pike.lysator.liu.se caudium.net societyserver.org > secretary beijinglug.org > mentor fossasia.org > foresight developer foresightlinux.org realss.com > unix sysadmin > Martin BÃ¤hr working in china http://societyserver.org/mbaehr/ >

Nov. 19, 2015

Array : very weird behavior

by abdelghani ALIDRA

Hi everybody, I am experiencing a very weird behavior with Arrays : I writed a test case as follows ; testChange | bitArray1 | bitArray1 := #(0 0 0 0). (1 to: bitArray1 size) do: [ :index ||temp| temp := (bitArray1 at: index) + 1. bitArray1 at: index put: (temp). ]. self assert: bitArray1 = #(1 1 1 1) The thing is that the test succedes the first time but if I run it again it fails.Moreover, if Â I change theÂ testChangeÂ arbitrarily (just add a space and save) the test succeeds the first time then it fails.....etc. ActuallyÂ bitArray1 := #(0 0 0 0)Â changes the value of bitArray1 in the first run but then bitArray1 does not change any more in the next runs. The problem does not occur if Â I replaceÂ bitArray1 := #(0 0 0 0). withÂ bitArray1 := Array new:4 withAll: 0 . Any idea where the problem can be?Should I open a bug entry? where? Abdelghani

Nov. 19, 2015

More fun with Roassal: 1st iteration of a Roassal Mindmapping tool

by Martin BÃ¤hr

Excerpts from stepharo's message of 2015-08-22 09:55:04 +0200: > would be fun to see if you can reuse the algo of roassal. i have taken a stab at that: =========================================================================================== view := RTView new. | activenode prevnode | view when: TRMouseClick do: [ :ev | | label node | "node := shape element." label := RTEditableLabel new height: 30; "i have a very high resolution and like large fonts" text: ''; element. node := RTBox new color: Color white; borderWidth: 2; borderColor: Color black; element. node @ RTDraggable. "select a node to branch of new nodes from it" node when: TRMouseClick do: [ :click | activenode := click element ]. "if the label is empty: '', then its size is 0@0 and is no longer clickable to edit, hence we make the node clickable instead" node when: TRMouseDoubleClick do: [ label shape showTextMorphFor: label ]. node when: TRMouseLeave do: [ label shape hideTextMorphFor: label ]. view add: node; add: label. RTNest new stretchable; on: node nest: {label}. "when the label changes its size, the box doesn't update its size unless we add this workaround" label trachelShape addCallback: (TRExtentCallback new block: [ :aShape :ratio | node trachelShape extent: aShape extent + (20 @ 0) ]). node translateTo: ev position. "label shape showTextMorphFor: label." "this opens the node in edit mode, but somehow messes things up" (activenode notNil and: node ~= activenode) ifTrue: [ prevnode := activenode ]. prevnode ifNotNil: [ edge := RTEdge from: prevnode to: node. edge + (RTBezierLine horizontal withContinuousRectangleAttachPoint; yourself). view add: edge ]. prevnode := node. activenode := node. view signalUpdate ]. view. =========================================================================================== the behavior is a bit different than the usual mindmapping tools as it does not enforce an automatic layout but you are free to place nodes whereever you like by just clicking in the location and moving the node later. that behavior is inspired by a mindmapping tool called labyrint: https://github.com/labyrinth-team/labyrinth https://www.youtube.com/watch?v=Pv3dTovGc0E thanks to alexandre bergel and others who helped me to figure out some problems such as a workaround for RTEditableLabel not working with RTCompositeShape: shape := RTCompositeShape new. shape allOfSameSizeWithPadding. shape add: (RTBox new color: Color white; borderWidth: 2; borderColor: Color black). shape add: (RTEditableLabel new height: 30). the remaining problems are cosmetic. for example i can't seem to get the nodes to open up in edit mode. doing that with the commented line in the code above, wrecks havoc on the node. i also would like RTBezierLine to be horizontal or vertical depending on how the two nodes are positioned to each other. and it should always start at the middle of the rectangles edge at a 90degree angle. lastly, i'd like to have rounded corners on the RTBox. if/when i develop this further, i plan to add the ability to select multiple nodes and add edges between them. be able to move selected nodes as a group, and align selected (or all) nodes according to one of the layouts provided by roassal (tree, circle, etc), things like export to svg or png would also be nice. and of course, package it up... greetings, martin. -- eKita - the online platform for your entire academic life -- chief engineer eKita.co pike programmer pike.lysator.liu.se caudium.net societyserver.org secretary beijinglug.org mentor fossasia.org foresight developer foresightlinux.org realss.com unix sysadmin Martin BÃ¤hr working in china http://societyserver.org/mbaehr/

Nov. 19, 2015