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

Re: [Pharo-project] Place to sleep Lille?

by StÃ©phane Ducasse

There are a couple of hotel: opera Agora Hotel 14 Rue du Molinel, 59000 Lille 03 20 31 55 52 â Â· agorahotel.com st maurice HÃ´tel Saint-Maurice 8 Parvis St Maurice, 59800 Lille 03 20 06 27 40 â Â· hotelsaintmaurice.com 1 HÃ´tel de la Paix 46 Bis r Paris 59800 Lille 03 20 54 63 93 Hotel du moulin d'or â¢ 03 66 88 05 65 15 r Molinel 59800 Lille else you also have formule 1 (mons en baroeul). av SabliÃ¨re 59370 MONS EN BAROEUL this is close to my place (three stops from flandres) 08 91 70 53 18 let us know Stef > Try this one: > > Hotel ibis Lille OpÃ©ra > 21 Rue Lepelletier, 59000 Lille, France > +33 3 20 06 21 95 > > -- > Damien Cassou > http://damiencassou.seasidehosting.st > > "Lambdas are relegated to relative obscurity until Java makes them > popular by not having them." James Iry >

May 21, 2012

Re: [Pharo-project] Browsing long methods

by StÃ©phane Ducasse

Yes but if we do not improve the system and gaucho is not usable by other than you then at the end we got nothing so the solution is to improve the system and eventually be ready to switch to something better but not before :) Stef > PluggableTextMorph and friends are a mess, i've tried to clean it up > once and found it to be really hard. > > I will be showing the support for text in Shapes ( the UI framework > used by Gaucho), which lays out text using Pango, and renderers it > using Athens (with a Cairo backend). Simple and yet powerful, thanks > to Pango. > > Fernando > > > On Mon, May 21, 2012 at 12:42 PM, Guillermo Polito > <guillermopolito(a)gmail.com> wrote: >> Nice, the paragraph is calculated about 7~8 times in a pluggable text morph (depending on the user). >> >> And since the composition of a paragraph for a long method takes long (some minutes), it takes 7 times to render it (I've never was able to wait for it. I always lose my patience before). >> >> Guille >> >> On Mon, May 21, 2012 at 11:41 AM, Guillermo Polito <guillermopolito(a)gmail.com<mailto:guillermopolito@gmail.com>> wrote: >> Browsing a long method (lets say, of ~270KB of sourcecode) just hangs the image. >> >> Interrupting doesn't even give me control over my image again. It does no open a debugger, but info is printed in stdout, so I assume that my UIProcess died, but the event fetcher is still alive... >> >> One of the processes that get interrupted prints: >> >> SmallIdentityDictionary(SmallDictionary)>>at:ifAbsent: >> MorphExtension>>valueOfProperty:ifAbsent: >> MorphExtension>>valueOfProperty: >> ScrollBar(Morph)>>valueOfProperty: >> ScrollBar(Morph)>>theme >> ScrollBar>>normalThumbFillStyle >> ScrollBar>>adoptGradientColor: >> ScrollBar>>sliderColor: >> ScrollBar>>adoptPaneColor: >> PluggableMultiColumnListMorph(Morph)>>adoptPaneColor: in Block: [:m | m adoptPaneColor: paneColor] >> >> So I just changed #adoptPaneColor: in ScrollBar to do nothing, and I am able to interrupt again. So it makes me think here it is where the UIProcess gets killed... >> >> Now, for the browser stuff, it seems that TextComposer>>#composeAllLines never finishes. I'm still investigating, but any clue is helpful :). >> >> Guille >> >

May 21, 2012

Re: [Pharo-project] SystemNavigationTest>>setUp looks quite strange to me

by StÃ©phane Ducasse

>> >> >> The problem is that somehow the test is not silent. Executing that test >> without that makes announcements to be raised with fake classes created on >> test time, and people subscribed to those announcements break. For example, >> Nautilus+Rpackage are not expecting classes that are not packaged with >> RPackage. >> > > I don't know anything of Rpackage, nor browsed any code, but it sounds > like a requirement for Rpackage : > being robust to unpackaged clases/methods. Yes but the problem is not there. Nautilus got event that it should not get and it is checking that a package is created in RPackageOrganizer (but the test does it on purpose on an empty organizer). So we identified the problem and we will look at it after 15h30.

May 21, 2012

Re: [Pharo-project] [Pharo-consortium] We need your Pharo successes

by StÃ©phane Ducasse

thanks! On May 21, 2012, at 1:13 PM, garduino wrote: > http://forum.world.st/file/n4631140/ArduinoSoftwareWebPostAutomation.pdf > ArduinoSoftwareWebPostAutomation.pdf > > ok, check it now. > > Cheers. > > -- > View this message in context: http://forum.world.st/We-need-your-Pharo-successes-tp4628432p4631140.html > Sent from the Pharo Smalltalk mailing list archive at Nabble.com. >

May 21, 2012

Re: [Pharo-project] Place to sleep Lille?

by Stephan Eggermont

Damien wrote: >Try this one: > >Hotel ibis Lille OpÃ©ra >21 Rue Lepelletier, 59000 Lille, France >+33 3 20 06 21 95 Thank you. I already did. No space. Stephan

May 21, 2012

Re: [Pharo-project] Browsing long methods

by Fernando Olivero

PluggableTextMorph and friends are a mess, i've tried to clean it up once and found it to be really hard. I will be showing the support for text in Shapes ( the UI framework used by Gaucho), which lays out text using Pango, and renderers it using Athens (with a Cairo backend). Simple and yet powerful, thanks to Pango. Fernando On Mon, May 21, 2012 at 12:42 PM, Guillermo Polito <guillermopolito(a)gmail.com> wrote: > Nice, the paragraph is calculated about 7~8 times in a pluggable text morph (depending on the user). > > And since the composition of a paragraph for a long method takes long (some minutes), it takes 7 times to render it (I've never was able to wait for it. Â I always lose my patience before). > > Guille > > On Mon, May 21, 2012 at 11:41 AM, Guillermo Polito <guillermopolito(a)gmail.com<mailto:guillermopolito@gmail.com>> wrote: > Browsing a long method (lets say, of ~270KB of sourcecode) just hangs the image. > > Interrupting doesn't even give me control over my image again. Â It does no open a debugger, but info is printed in stdout, so I assume that my UIProcess died, but the event fetcher is still alive... > > One of the processes that get interrupted prints: > > SmallIdentityDictionary(SmallDictionary)>>at:ifAbsent: > MorphExtension>>valueOfProperty:ifAbsent: > MorphExtension>>valueOfProperty: > ScrollBar(Morph)>>valueOfProperty: > ScrollBar(Morph)>>theme > ScrollBar>>normalThumbFillStyle > ScrollBar>>adoptGradientColor: > ScrollBar>>sliderColor: > ScrollBar>>adoptPaneColor: > PluggableMultiColumnListMorph(Morph)>>adoptPaneColor: in Block: [:m | m adoptPaneColor: paneColor] > > So I just changed #adoptPaneColor: in ScrollBar to do nothing, and I am able to interrupt again. So it makes me think here it is where the UIProcess gets killed... > > Now, for the browser stuff, it seems that TextComposer>>#composeAllLines never finishes. Â I'm still investigating, but any clue is helpful :). > > Guille >

May 21, 2012

Re: [Pharo-project] Place to sleep Lille?

by Damien Cassou

Try this one: Hotel ibis Lille OpÃ©ra 21 Rue Lepelletier, 59000 Lille, France +33 3 20 06 21 95 -- Damien Cassou http://damiencassou.seasidehosting.st "Lambdas are relegated to relative obscurity until Java makes them popular by not having them." James Iry

May 21, 2012

Re: [Pharo-project] Morpheas , bringing Morphic to Opengl (3d GUIs)

by Javier PimÃ¡s

Could you put a self halt in NBGLXContextDriver>>createContext: and step over line by line to see which one is the offending one? On Mon, May 21, 2012 at 12:55 AM, chadwick <mathnoir(a)gmail.com> wrote: > more detail: after I do GLTTRenderingDemo new openInWorld, I tried > dragging a window around in a circle so as to get a feel for what the CPU > is doing. The image pauses for increasingly longer times, making the UI > unresponsive during this time, eventually locking up entirely. > -- Lic. Javier PimÃ¡s Ciudad de Buenos Aires

May 21, 2012

Re: [Pharo-project] WTF? why this can happen?

by Esteban Lorenzano

yeah... solved... thanks :P On May 21, 2012, at 1:39 PM, Nicolas Cellier wrote: > Maybe you need HashedCollection rehashAll ? > > Nicolas > > 2012/5/21 Esteban Lorenzano <estebanlm(a)gmail.com>: >> Hi, >> >> "expanded" is a Set with just one element. >> >> executing (somewhere inside pier): >> >> expanded removeAll: expanded copy >> >> throws NotFound >> >> digging a bit, I found that the #scanFor: message answers 5 and real position of object in array is 1 (array size is 5, btw). >> >> in: >> >> scanFor: anObject >> "Scan the key array for the first slot containing either a nil (indicating an empty slot) or an element that matches anObject. Answer the index of that slot or raise an error if no slot is found. This method will be overridden in various subclasses that have different interpretations for matching elements." >> >> | index start | >> index := start := anObject hash \\ array size + 1. >> [ >> | element | >> ((element := array at: index) == nil or: [ element enclosedSetElement = anObject ]) >> ifTrue: [ ^index ]. >> (index := index \\ array size + 1) = start ] whileFalse. >> self errorNoFreeSpace >> >> >> this line: >> >> index := start := anObject hash \\ array size + 1. >> >> is answering 5... so method ends on first ifTrue: >> >> but... what the hell???????? >> any idea? >> >> thanks, >> Esteban >

May 21, 2012

Re: [Pharo-project] SystemNavigationTest>>setUp looks quite strange to me

by Nicolas Cellier

2012/5/21 Guillermo Polito <guillermopolito(a)gmail.com>: > > > On Mon, May 21, 2012 at 7:57 AM, StÃ©phane Ducasse > <stephane.ducasse(a)inria.fr> wrote: >> >> Hi guillermo >> >> I saw that you modified >> >> setUp >> Â Â Â Â super setUp. >> >> Â Â Â Â oldMonticelloAnnouncer := SystemAnnouncer privateAnnouncer. >> Â Â Â Â SystemAnnouncer instVarNamed: 'privateAnnouncer' put: nil. >> >> Â Â Â Â navigator := self createSystemNavigationToTest. >> Â Â Â Â classFactory := self createClassFactory. >> >> >> Now I would like to understand why? Especially because privateAnnouncer >> should not be used and touched. > > > The problem is that somehow the test is not silent.Â Executing that test > without that makes announcements to be raised with fake classes created on > test time, and people subscribed to those announcements break.Â For example, > Nautilus+Rpackage are not expecting classes that are not packaged with > RPackage. > I don't know anything of Rpackage, nor browsed any code, but it sounds like a requirement for Rpackage : being robust to unpackaged clases/methods. Nicolas >> >> SystemAnnouncer privateAnnouncer returns an announcer that should only be >> used by internal (not tools). > > > Actually, it's just a patch to make the tests run.Â But the real problem is > the non-silentlyness of the tests. > >> >> >> BTW the teardown looks wrong to me. > > > Yeap, I got it wrong there.Â But now I do not know if to fix the patch or > attack the real issue is the way to go :).Â Look at my comment in the issue > tracker. > > Guille

May 21, 2012