July 2011 - Pharo-dev - lists.pharo.org

Re: [Pharo-project] [Vm-dev] which vm on 64bit Linux?

by Nick Ager

On 19 July 2011 15:52, Mariano Martinez Peck <marianopeck(a)gmail.com> wrote: > Anyone. As if it were 32 bits. The only difference is that you may need to > install the 32bits library for the OS. > If you say which linux someone may help you how to install such libs. > On Centos you I install: sudo yum install xauth mesa-libGL.i386 libXrender.i386 libSM.i386 freetype.i386 libstdc++.i386 on Ubuntu: sudo aptitude install ia32-libs

July 19, 2011

[Pharo-project] Exception >> return:?

by Camillo Bruni

Can anyone explain me why it would make sense to use #return: in the following method? TestCase >> executeShould: aBlock inScopeOf: anExceptionalEvent ^[aBlock value. false] on: anExceptionalEvent do: [:ex | ex return: true] I would instinctively replace [:ex | ex return: true] with the much simpler [:ex | ^ true].. Am I missing something here? camillo

July 19, 2011

Re: [Pharo-project] why isEmptyOrNil is not defined in Object?

by StÃ©phane Ducasse

On Jul 19, 2011, at 12:08 PM, Lukas Renggli wrote: > I would say for the same reason #isEmpty is not defined on Object, but > only on Collection. #isEmptyOrNil is the equivalent of > > obj isNil or: [ obj isEmpty ] yes this is what I did. I do not see the point of isEmptyOrNil just because people returns nil or '' while a default decent value should be returned. > (which i find clearer and) which doesn't work on arbitrary objects either. > > Lukas > > On Tuesday, 19 July 2011, StÃ©phane Ducasse <stephane.ducasse(a)inria.fr> wrote: >> Hi guys >> >> I'm puzzled and probably need food. but why isEmptyOrNil is not defined in Object? >> because right now if a variable is '' or nil or something else I cannot use it. >> >> Stef >> > > -- > Lukas Renggli > www.lukas-renggli.ch >

July 19, 2011

Re: [Pharo-project] How Moprhs is rendered

by StÃ©phane Ducasse

On Jul 19, 2011, at 12:46 PM, Hilaire Fernandes wrote: > To optimise the DrGeo canvas (get it faster), I try to understand in > detail how is done the rendering of Morphs embedded in a pasteupmorph. > I did not find any detailed information on the Internet about that. > Any pointer? not that I know but I would be interested too. > > > Right now, DrGeo canvas is a pasteUpMorph, the line, segment, point, > circle, etc. are all Morph added in the pasteUpMoprh. When the model of > a geometry sketch is updated, then all the Morphs view are updated in > one shot (it appeared to be faster than to climb the model tree to > update only the morphs needing update) > > Hilaire > > -- > Education 0.2 -- http://blog.ofset.org/hilaire > >

July 19, 2011

Re: [Pharo-project] [Vm-dev] which vm on 64bit Linux?

by Mariano Martinez Peck

Anyone. As if it were 32 bits. The only difference is that you may need to install the 32bits library for the OS. If you say which linux someone may help you how to install such libs. cheers On Tue, Jul 19, 2011 at 4:49 PM, Tudor Girba <tudor(a)tudorgirba.com> wrote: > > Hi, > > Which VM should be used to run Pharo on a 64bit Linux? > > Cheers, > Doru > > > -- > www.tudorgirba.com > > "Every successful trip needs a suitable vehicle." > > > > > -- Mariano http://marianopeck.wordpress.com

July 19, 2011

[Pharo-project] which vm on 64bit Linux?

by Tudor Girba

Hi, Which VM should be used to run Pharo on a 64bit Linux? Cheers, Doru -- www.tudorgirba.com "Every successful trip needs a suitable vehicle."

July 19, 2011

Re: [Pharo-project] why isEmptyOrNil is not defined in Object?

by Nicolas Cellier

"deprecate it" was a bit provocative, and would be hard to achieve now. But you agree that we gratuitously complexified a legacy simple interface, while there were smoother, and IMO more elegant solutions, don't you? Nicolas 2011/7/19 Esteban Lorenzano <estebanlm(a)gmail.com> > well... I use it a lot... yes, mainly for "FillInTheBlank". > In fact, I usually add it to my programs... > however, I'm not sure if we have to deprecate it (and replace for a larger > code), or adopt it :) > > cheers, > Esteban > > El 19/07/2011, a las 7:14a.m., Nicolas Cellier escribiÃ³: > > I don't like isEmptyOrNil. > It has been introduced mainly because we can now distinguish these two > operations in a "FillInTheBlank" > - cancel returns nil > - accept returns a String eventually empty > The original implementation did always return an empty String even if > cancel was pressed. > > But no application is really prepared to deal with nil, and most did > interpret an empty String as a cancel. > So we replaced the isEmpty checks by isEmptyOrNil... > > I would have done it differently: > 1) let the default return an empty String even if cancelled > 2) provide a ifCancelled: addtional key for the rare apps which really need > to distinguish. > > UIManager default request: 'OK ?' initialAnswer: '' ifCancelled: [^nil]. > > Default behaviour would be equivalent to > UIManager default request: 'OK ?' initialAnswer: '' ifCancelled: ['']. > > So my take would be to deprecate it. > > Nicolas > > 2011/7/19 StÃ©phane Ducasse <stephane.ducasse(a)inria.fr> > >> Hi guys >> >> I'm puzzled and probably need food. but why isEmptyOrNil is not defined >> in Object? >> because right now if a variable is '' or nil or something else I cannot >> use it. >> >> Stef >> > > >

July 19, 2011

Re: [Pharo-project] fun bug: fileouting method with a $!, produces $!!

by Nicolas Cellier

2011/7/18 StÃ©phane Ducasse <stephane.ducasse(a)inria.fr> > But I cannot recompile the method I fileout. > Can you? > May be this is because I did not use the filelist and I copied and pasted > from the file veronica sent me. > > An alternative format not using escape sequence would be to define the end of chunk on line, like in a shell script: cat > myfile << END here is my own data END The chunk could be chosen by default and changed only if present in method source (or class comment or whatever). Particularly, if you want to transcribe class comments unaltered (that is without quoting of any form), you can't rely on any syntax to delimit the comments, there is none. > > this is not because people were not using a decent syntax to fileout and > that after they hacked the system that this is not a bug. > > Yes, it is by no way meant to be human written and does not decently meet > the requirements for scripting. > > For exchanging code, the chunk format is a bit poor too, but it is > manageable with a good change list tool (not the squeak one). > > And remember, it was (and is still) used for logging system changes in > the change log. > > And it is very efficient for that. > > With chunk format, you don't need quoting the code (except exclamation > mark itself of course), and that's an important feature > > IMO, this format served its purpose very decently. > > Well, you can tell me the Squeak hacks to identify authors and timestamp > are awfully implemented, I can only agree, but the basis are very simple. > TSTTCPW. > > I think that it would deserve a new pass. Because "very simple" is not true > when you look at the complexity = invoking the scanner to get some > information and then poking around to know if this is a comment or a class > definition. > > Yes. The format is mainly like this: 1) metadata 2) source If metadata is missing or incomplete, then the tools have to try to decode the source (generally a do-it). This is particularly true in Squeak for a variety of actions (I don't have an image handy, but I think removing renaming a class etc... are in this category). > > But you can. > > You can browse the code, can't you ? go and see in .sources or .changes, > you'll see a doubled ! > > You can fileOut, then fileIn, no problem. > > I spent some days a couple of year ago in the filein fileout because I > wanted to load all the code of squeak starting from sq10 -> 3.9 and > I can tell you that the allButLast first .... kind of expressions to skip > over ' or ! to me indicate a problem because saving a element of program > should not be like that and this is why the format of the fileout is not > good. > Of course it works but it does not mean that this is good. > > Stef > Totally agree, as I said before, I don't like the implementation because it hardly readable, maintainable, and is very inefficient... Especially the handling of Squeak meta additions (timestamp, author). But the format by itself is very simple. I don't buy the terms "good" as an absolute reference, it must be relative to the requirements. Expose your requirements, then we'll can say which format is good or not. Nicolas

July 19, 2011

Re: [Pharo-project] why isEmptyOrNil is not defined in Object?

by Esteban Lorenzano

well... I use it a lot... yes, mainly for "FillInTheBlank". In fact, I usually add it to my programs... however, I'm not sure if we have to deprecate it (and replace for a larger code), or adopt it :) cheers, Esteban El 19/07/2011, a las 7:14a.m., Nicolas Cellier escribiÃ³: > I don't like isEmptyOrNil. > It has been introduced mainly because we can now distinguish these two operations in a "FillInTheBlank" > - cancel returns nil > - accept returns a String eventually empty > The original implementation did always return an empty String even if cancel was pressed. > > But no application is really prepared to deal with nil, and most did interpret an empty String as a cancel. > So we replaced the isEmpty checks by isEmptyOrNil... > > I would have done it differently: > 1) let the default return an empty String even if cancelled > 2) provide a ifCancelled: addtional key for the rare apps which really need to distinguish. > > UIManager default request: 'OK ?' initialAnswer: '' ifCancelled: [^nil]. > > Default behaviour would be equivalent to > UIManager default request: 'OK ?' initialAnswer: '' ifCancelled: ['']. > > So my take would be to deprecate it. > > Nicolas > > 2011/7/19 StÃ©phane Ducasse <stephane.ducasse(a)inria.fr> > Hi guys > > I'm puzzled and probably need food. but why isEmptyOrNil is not defined in Object? > because right now if a variable is '' or nil or something else I cannot use it. > > Stef >

July 19, 2011

Re: [Pharo-project] [update 1.4] #14050

by Pavel Krivanek

Hi Stef, is there some paper about Ring? Should it be part of the kernel image? -- Pavel On Tue, Jul 19, 2011 at 1:44 PM, StÃ©phane Ducasse <stephane.ducasse(a)inria.fr> wrote: > > > 14050 > ----- > > Preparation for Ring integration > >

July 19, 2011