November 2008 - Pharo-dev - lists.pharo.org

[Pharo-project] adrian .... help :)
by StÃ©phane Ducasse Nov. 8, 2008

Nov. 8, 2008

adrian could you have a look at the scriptloader because it gets desynchronised (I had to change the number by hand this morning again and I followed exactly the process). May be my image counter was not correct. I think that this counter is brittle. I would just do what we did before taking the lastest updateXXX number + 1 Stef

2 3

[Pharo-project] About Pharo - text to go with version 1.0
by Matthias Berth Nov. 8, 2008

Nov. 8, 2008

Hello, I am thinking that some "getting started" info is a good idea for Pharo 1.0. As a first part, I have tried to compile various statements "about pharo" and what I read on the mailing list into a short text (see below). Does that capture what people doing the work are thinking about the project? Anything I should add? Anything you don't like? Cheers Matthias =About Pharo= The goal of the Pharo Project is to produce a clean and lean open-source Smalltalk for professional software development: * great tools, e.g. OmniBrowser with refactoring support * polished user interface with TrueType fonts and Polymorph widgets * clean and lean image * works out-of-the-box, with clear ways to add functionality Starting from version 1.0, Pharo wants to be stable to make sure that we can produce ideas based on it. An important test for Pharo is that mission critical Seaside (http://www.seaside.st) applications should run on it. Pharo started from Squeak (http://www.squeak.org) [which version, 3.10?] but it is a fork: Do not expect Etoy/Nebraska to work in Pharo. We will try to avoid to reinvent the wheel but will change if changing is better. Pharo wants to take a fresh look at the Smalltalk philosophy and current implementations. Ideally by 2011 we would like to have a small kernel and some cool packages that we can load on top. The current focus is not to produce such a kernel but good packages. Everybody is welcome to contribute. You can find a wiki, bug tracker, and a mailing list via the Pharo home page at http://www.pharoproject.org ==License== [A few words about the code that is already there in Pharo, please] Any code developed and committed to the SqueakSource/Pharo and PharoInbox repository is by default under the MIT License.

3 4

Fwd: [Pharo-project] Sunit Extensions breaking the TestRunner
by StÃ©phane Ducasse Nov. 8, 2008

Nov. 8, 2008

Begin forwarded message: > From: Keith Hodges <keith_hodges(a)yahoo.co.uk> > Date: November 8, 2008 2:18:32 AM CEST > To: Alexandre Bergel <Alexandre.Bergel(a)inria.fr> > Cc: stephane ducasse <stephane.ducasse(a)free.fr> > Subject: Re: [Pharo-project] Sunit Extensions breaking the TestRunner > > Alexandre Bergel wrote: >> The situation is getting messy. >> >> Cheers, >> Alexandre > exactly... everyone doing their own thing for their own reasons. Will > you be willing to help restore some coherence to the picture. > > I set a few goals that are fulfilled in my derived implementation. My > improvements are based upon more than 10 years of experience of using > SUnit and things like it. These have been out there for 2 years, so > there is no reason why people cannot build upon this base. > > 1. Ability to define test suites, and have a UI for running test > suites. > I have been using this facility for 8 years in st/x and always miss it > in squeak. I want to be able to define a suite for "My Project". > > 2. Ability to flag which tests are expected to work in which > situation, > which image version and which platform. So that I can run with "all > tests green" even if I happen to have a few windows specific tests, > when > I am running on a mac. > > 3. Ability to use tests to specify behaviour and requirements even > before the tests work. I.e. mark tests as #toDo. > > 4. Ability to run tests in an automated tool, without a UI. > > 5. Ability to sort tests so that they are run fastest first. 95% of > the > tests will run in 10% of the time. > > 6. Ability to determine whether a test used the network or not, and > filter out those tests if necessary. > > 7. Ability to integrate other testing technologies like SSpec. (not > actually done, but conceptually possible now) > > A couple of other improvements are also included, such as the > ability to > define a custom exception handler around all of your tests. > > In my opinion some members of the Pharo team have a big problem with > "Not Invented Here". They don't even look at the existing stuff before > diving in. > > I spent several months integrating the fixes from 3 separate streams > of > Monticello, specifically with the goal of promoting coherence within > all > of the the squeak communities. I did this work for the benefit of all > squeak users, not for the benefit of my health or financial well > being. > In fact it has cost me several thousands of pounds in lost revenue. > > I am actually fairly insulted by the attitude of the Pharo team > towards > this work. Since they take MC and continue to work on it > separately, it > actually renders all of my effort a waste of time. > > perhaps we can rescue sunit before it is too late. > > Keith > >

1 1

[Pharo-project] Strange primitives doing nothing
by Igor Stasenko Nov. 7, 2008

Nov. 7, 2008

I'm exploring the exception support in VM, and found two strange primitives used by BlockClosure, which simply fail and doing nothing. primitiveMarkUnwindMethod "Primitive. Mark the method for exception unwinding. The primitive must fail after marking the context so that the regular code is run." self inline: false. ^self primitiveFail primitiveMarkHandlerMethod "Primitive. Mark the method for exception handling. The primitive must fail after marking the context so that the regular code is run." self inline: false. ^self primitiveFail Any ideas, why they still in use? Can there be something in between, when interpreter enters the method? Or, if nothing, then why they still there? :) -- Best regards, Igor Stasenko AKA sig.

1 0

[Pharo-project] [ann] Pharo0.1Core-10133.zip
by Marcus Denker Nov. 7, 2008

Nov. 7, 2008

https://gforge.inria.fr/frs/download.php/10877/Pharo0.1Core-10133.zip ChangeLog: ========== ------------------------------------------------- update10108 - removing some unsent methods from etoy s to reduce the fat. ------------------------------------------------------ update10109 - removing etoy -> ClassDescription - isUniClass senders and implementors ------------------------------------------------------ update10110 Issue 284: add literalsDo: as deprecated method Issue 253: Mantis 6933 MultiByteFileStream upTo: does not work in binary mode Issue 252: Mantis 6812 cannot inspect a WeakSet - add back deprecated #literalsDo: ------------------------------------------------------ update10111 - Debugger etoy removal - remove etoyFriendly preference - red/green for conflicts in MC (damien Pollet) ------------------------------------------------------ update10112 Issue 260: TODO: cleanup hasLiteralThorough: ------------------------------------------------------ update10113 - FileDirectory etoy cleaning - Project / ThreadNavigation method cleaning ------------------------------------------------------ update10114 - Issue 274: TwoWayScrollPane shoud be removed - Issue 291: SystemNavigation tests + reimplementation for queries on sent and unsent messages - Issue 290: SUnit Enhancement to ease writing tests involving the creation of classes ------------------------------------------------------ update10115 - Issue 274: TwoWayScrollPane shoud be removed - Issue 291: SystemNavigation tests + reimplementation for queries on sent and unsent messages - Issue 290: SUnit Enhancement to ease writing tests involving the creation of classes ------------------------------------------------------ update10116 - Issue 185: More World menu cleanup needed (includes the following improvements: mouse-over sub-menus, capitalize tool names, exchange single with double change sorter) - Issue 282: Increase default window sizes - Issue 283: Remove fonts and change default font choices - ScriptLoader>>cleanUpForRelease: flushes a lot of caches and deletes bitmaps from EToys to decrease the memory footpring ------------------------------------------------------ update10117 - Issue 293: SystemNavigation UI + bug fixes - Installer installer in scriptloader updated - Issue 297: MethodReference printOn enhance - Etoy unsentMessages removed!!!!! :) ------------------------------------------------------ update10118 - fix script loading broken. Issue 294: VM does not send script files to image ------------------------------------------------------ update10119 - fix script loading broken. Issue 294: VM does not send script files to image ------------------------------------------------------ update10120 - Etoy removal TextMorph ThumbnailMorph TileMorphTest TraitDescription UndefinedObjectTest UpdatingStringMorph Vocabulary ------------------------------------------------------ update10121 - Etoy clean up FileList2 and other little left over ------------------------------------------------------ update10122 - Merging Type Hierarchy removal second attempt ------------------------------------------------------ update10123 - Stef and David Hacking on Morph etoy Removal - moved ScriptEditorMorph -> Etoy ------------------------------------------------------ update10124 - Revert to Accuny as the default font and remove the other font from the system ------------------------------------------------------ update10125 Issue 303: TODO: remove profiling methods Issue 300: unimplemented call in AlignmentMorphBob1>>acceptDroppingMorph:event: Issue 299: Mantis 007007 and 007008 correct toggle break on entry ------------------------------------------------------ update10126 -> Issue 248: SLICE-Mantis-ScaledDecimal-SqNumberParser-Fix-nice.1 -> Issue 295: #testSupplyAnswerUsingRegexMatchOfQuestion fails -> latest collection tests ------------------------------------------------------ update10129 - delete unused vars in Morphic (Lukas R.) - fix loop of liked list test (Damien P.) ------------------------------------------------------ update10129 Various etoy cleaning ------------------------------------------------------ update10130 313: FileList2 crashes if some subdirectory present ------------------------------------------------------ update10132 Fix some tests (duplicated methods from traits) ------------------------------------------------------ update10132 Removed ccg:emitLoadFor:from:on: ccg:generateCoerceToOopFrom:on: ccg:generateCoerceToValueFrom:on: ccg:prolog:expr:index: ============== (full list of changes: print "ScriptLoader new logContents ' -- Marcus Denker -- denker(a)iam.unibe.ch http://www.iam.unibe.ch/~denker

4 4

[Pharo-project] Strange bugs?
by StÃ©phane Ducasse Nov. 7, 2008

Nov. 7, 2008

Hi guys I would be interesting to understand these two points: - times repeat changes - hex removal (I do not remember why we removed it). Stef Begin forwarded message: > From: lists(a)moedae.net > Date: November 2, 2008 12:36:59 AM CEST > To: stephane ducasse <stephane.ducasse(a)free.fr> > Subject: Re: [Seaside] how much introduction? > Reply-To: taun <lists(a)moedae.net> > > More details about the timesRepeat and hex problems I had. I know > when I found them I asked myself why someone would have touched > timesRepeat and why hex would have been removed. Neither made any > sense. Yet I also know my first view of each method in the image I > was using was through the debugger after the problem had become > apparent. > > timesRepeat background. > > Towards the beginning of this week, > I downloaded the latest Pharo image. > Opened and updated the image. > Saved image. > Used Monticello to loaded the application I was working on. > Application had some OrderedCollections which were missing the last > element value. > Checked my loop limits and found them to be correct. > Saw it was runnning my loop once rather than twice and this was > happening at timesRepeat:. > Debugged and stepped through including stepping into the > timesRepeat: method. > Saw timesRepeat had < rather than <=. > Scratched my head wondering why on the world this would be. > I checked the versions to see if there were initials and a previous > version. There weren't > Updated image. > Ran my application and it was fine. > Checked timesRepeat: and it was back to normal. > Checked my code in Monticello and it had no overrides for > timesRepeat:. > > A mystery. One guess would be in stepping through the code, the > delete key was hit while in the timesRepeat: method and the method > was saved. How I could delete, save then have to restart the > debugging all without realizing it is also a mystery. I have two > young kids so maybe one of them helped with the distraction. > > > > As for the missing Color #hex:, I looked through the Pharo change > log and found,... > > !ScriptLoader methodsFor: 'sapphire - scripts' stamp: > 'stephane.ducasse 6/8/2008 22:28'! > script124 > > | names| > names := '39Deprecated-sd.12.mcz > ... > Cut lots of code from here but see below. > ... > Color class removeSelector: #hex:! > > "Graphics"! > > "GraphicsTests"! > > "Tests"! > > "Postscript: > Leave the line above, and replace the rest of this comment by a > useful one. > Executable statements should follow this comment, and should > be separated by periods, with no exclamation points (!!). > Be sure to put any further comments in double-quotes, like this one." > > |repository| > repository := MCHttpRepository > location: 'http://www.squeaksource.com/Pharo/' > user: '' > password: ''. > (repository loadVersionFromFileNamed:'ScriptLoader-stephane.ducasse. > 434.mcz') load. > ScriptLoader new update10034. > ! > > ----End fileIn of a stream----! > > Taun > > ----- Original Message ----- > From: "stephane ducasse" <stephane.ducasse(a)free.fr> > To: "taun" <lists(a)moedae.net>, "Seaside - general discussion" <seaside(a)lists.squeakfoundation.org > > > Sent: Saturday, November 1, 2008 5:49:23 AM GMT -05:00 US/Canada > Eastern > Subject: Re: [Seaside] how much introduction? > > > On Nov 1, 2008, at 5:56 AM, lists(a)moedae.net wrote: > >> Having not done a survey, I can only talk from my personal >> experience. So far my personal experience is I have had about the >> same number of issues in 2 weeks of using Pharo as in 2 years with >> Squeak 3.9 and 3.10. >> >> In Pharo, some OmniBrowser functionality was broken and as I said >> timesRepeat: was broken. It had a '<' rather than '<=' for the loop >> comparison so the loop ran one too few times. I ran an update which >> fixed both problems. However the update also changed the World menu >> and eliminated the flaps (I like the new menu and the flaps were >> redundant). As someone who has used SmallTalk for over 20 years and >> someone who hopefully understands why Pharo is so dynamic at the >> moment, I can live with the issues mentioned previously. When I >> debugged my loop, got to the timesRepeat: method and saw the test >> condition, it was obvious what was wrong. > > This is strange since I did not recall a change at that level. > >> I then performed an update in the hope that someone had already >> fixed the problem. They had. I wanted to try out SmallFaces. I >> loaded and started to test SmallFaces in the latest Pharo image. The >> code broke. It turns out Pharo's Color class was missing #hex:. A >> method in the 3.9 > > This is strange since we started from 3.9 > >> and 3.10 images and used by SmallFaces. Again, a minor item to fix, >> but not if you are new to SmallTalk and Squeak/Pharo. > > I agree pharo is probably changing too fast for a new comers. > I should have say: 'Sergio if you hate the long menus of squeak have a > look at pharo. If one day you start to look at squeak and start to > think that there > are some really bad smell like core behavior depending on UI, then > have a look at pharo' > > Note please report to us any problem you get for running your > applications. > >> Before recommending the current version of Pharo to a new >> SmallTalker, I think one really needs to ask oneself how they would >> react if while feeling lost trying to learn a new language and >> environment, the menus for the environment suddenly changed after an >> update. Or their potential confusion when the environment doesn't >> match all the existing tutorials. > > I agree. > BTW: this is Smalltalk and not SmallTalk > >> I once worked at a company where we were growing so fast, every day >> I went to work, the office walls had moved. I started to have dreams >> about falling asleep at my desk and waking up to find my office door >> replaced by a wall. The growth was exciting because it was a sign of >> our success but it definitely made life difficult for new employees. >> This is Pharo at the moment. > > :) > Good.! > >> >> >>> I would argue the inverse: Take soon the one click image with the >>> new >>> browser of david and the clean menu. >> >> Yes, when it is released and stable it would be better. But it is >> neither at present. > > It depends what you call stable :) >> >> >> Taun >> >> ----- Original Message ----- >> From: "David Mitchell" <david.mitchell(a)gmail.com> >> To: "Seaside - general discussion" >> <seaside(a)lists.squeakfoundation.org> >> Sent: Friday, October 31, 2008 9:34:23 AM GMT -05:00 US/Canada >> Eastern >> Subject: Re: [Seaside] how much introduction? >> >> I agree with Stef, >> >> I moved my production code to Pharo last month and I've had no issue >> reports. >> >> I've loaded Seaside, MySql, and my own code. >> >> >> On Fri, Oct 31, 2008 at 7:36 AM, stephane ducasse >> <stephane.ducasse(a)free.fr> wrote: >>> >>> On Oct 29, 2008, at 3:17 PM, Taun wrote: >>> >>>> I think the real point Philippe made is a good one. It is not good >>>> for the >>>> Squeak/Smalltalk community to point a new user to an alpha >>>> product. The new >>>> user should be pointed to a stable functional platform so they can >>>> concentrate on learning the language not the platform bugs. >>> >>> Come on we are a good group of people using pharo daily and it has >>> less bugs >>> than squeak. >>> I know since I harvested the fixes. >>> >>>> Philippe's example of breakage was an example for why Pharo might >>>> not be >>>> good for Sergio. >>> >>> I think that this is not true. >>> Do you think that a newcomer is not overhlem by the squeak menu! >>> >>> >>>> As an additional example, I was using Pharo the other day and while >>>> testing one of my models, I discovered #timesRepeat: was broken. >>> >>> I would be really amazed that timesRepeat: would be broken in pharo. >>> Can you tell us, which version and how to reproduce that? >>> >>> >>>> Imagine Sergio running some simple sample code loops and getting >>>> the wrong >>>> answer. What is that going to do to his newly forming opinion of >>>> Squeak? >>>> >>>> Again, these are not criticisms of Pharo, simply reasons why Pharo >>>> might >>>> not be the right tool for a beginning Squeak/Seaside user. >>> >>> I would argue the inverse: Take soon the one click image with the >>> new >>> browser of david and the clean menu. >>> Pharo will have much more tests than any squeak distro. >>> Pharo contains far less broken code than 3.9 or 3.10. >>> >>> After the point is that we are working on it so when you update you >>> can get >>> some glitches. >>> But even with that I think that this is better than 3.9. >>> >>> Stef >>> >>>> >>>> >>>> Taun >>>> >>>> >>>> ----- Original Message ----- >>>> From: "Adrian Lienhard" <adi(a)netstyle.ch> >>>> To: "Seaside - general discussion" <seaside(a)lists.squeakfoundation.org >>>>> >>>> Sent: Monday, October 27, 2008 12:45:32 PM GMT -05:00 US/Canada >>>> Eastern >>>> Subject: Re: [Seaside] how much introduction? >>>> >>>> Pharo currently is in alpha, so its surprising you expect it to be >>>> free of bugs. We are making a lot of progress, so sometimes a >>>> change >>>> breaks some unexpected functionality. Usually, though, this is >>>> fixed >>>> within few days. We will soon go to beta and then produce a release >>>> that is going to be maintained. That's where you can expect >>>> stability. >>>> >>>> Adrian >>>> >>>> On Oct 27, 2008, at 17:25 , Philippe Marschall wrote: >>>> >>>>> 2008/10/27 stephane ducasse <stephane.ducasse(a)free.fr>: >>>>>> >>>>>> Hi sergio >>>>>> >>>>>> when you will be fit on the syntax and the environment. I suggest >>>>>> you to >>>>>> take >>>>>> the pharo image since it is really starting to get cleaner and >>>>>> cleaner. >>>>>> http://code.google.com/p/pharo/ >>>>> >>>>> I don't want to bash any of you but I have had some really bad >>>>> experiences with Pharo. Stuff like ClassBuilder exploding in my >>>>> face >>>>> and OmniBrowser not working. >>>>> >>>>> Cheers >>>>> Philippe >>>>> _______________________________________________ >>>>> seaside mailing list >>>>> seaside(a)lists.squeakfoundation.org >>>>> http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside >>>> >>>> _______________________________________________ >>>> seaside mailing list >>>> seaside(a)lists.squeakfoundation.org >>>> http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside >>>> _______________________________________________ >>>> seaside mailing list >>>> seaside(a)lists.squeakfoundation.org >>>> http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside >>>> >>> >>> _______________________________________________ >>> seaside mailing list >>> seaside(a)lists.squeakfoundation.org >>> http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside >>> >> _______________________________________________ >> seaside mailing list >> seaside(a)lists.squeakfoundation.org >> http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside >> _______________________________________________ >> seaside mailing list >> seaside(a)lists.squeakfoundation.org >> http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside >> > > >

6 11

[Pharo-project] Issues
by Torsten Bergmann Nov. 7, 2008

Nov. 7, 2008

>this is important when you open an issue to tell it in the mailing-list Hi Stef, http://code.google.com/p/pharo/issues/detail?id=312 can be harvested savely. It's a typo in a symbol There is also 314 ShoutWorkspace / Support for different Workspace types (adaptions if shout will make it into the std. image) 232 Recreate Special Objects Array should preserve external semaphores (the loading after FFI install problem) 231 Image does not open full screen anymore (outdated Win32 VM) Last one is obsolete when Markus uses newer sources for building the one click Win32 VM Bye Torsten -- GMX Download-Spiele: Preizsturz! Alle Puzzle-Spiele Deluxe Ã¼ber 60% billiger. http://games.entertainment.gmx.net/de/entertainment/games/download/puzzle/i…

2 1

Fwd: [Pharo-project] ArrayedCollection, Boolean SmallInteger>>ccg: cg generateCoerceToOopFrom: aNode on: aStream
by StÃ©phane Ducasse Nov. 7, 2008

Nov. 7, 2008

Begin forwarded message: > From: "David T. Lewis" <lewis(a)mail.msen.com> > Date: November 7, 2008 1:08:04 AM CEST > To: John M McIntosh <johnmci(a)smalltalkconsulting.com> > Cc: stephane.ducasse(a)inria.fr > Subject: Re: Fwd: [Pharo-project] ArrayedCollection, Boolean > SmallInteger>>ccg: cg generateCoerceToOopFrom: aNode on: aStream > > Hi John, > > It is safe to remove all the 'ccg:*' methods from Pharo, and no > other updates are required for VMMaker. > > All of the methods in question are already included in the VMMaker > package. The methods in the base image are exact duplicates of the > corresponding methods in the VMMaker package (except for method > category of course). I would guess that Tim may have just forgotten > to delete them. > > I have entered Mantis issue #7221 to track this, with a change set > to remove the duplicate methods from a base 3.10 image. I also > confirmed > that after removing the methods from the base image, I was able to > load > VMMaker and build a VM. > > - Dave > > > On Thu, Nov 06, 2008 at 12:34:11PM -0800, John M McIntosh wrote: >> David if you want to look into this that would be good. I think there >> are a few other methods (4 message sends I think) >> >> ccg:emitLoadFor:from:on: >> ccg:generateCoerceToOopFrom:on: >> ccg:generateCoerceToValueFrom:on: >> ccg:prolog:expr:index: >> >> on Array, Boolean, ByteArray, Float, FloatArray, IntegerArry, Object, >> and WordArray >> >> As the Pharo folks noticed these really should be part of VMMaker and >> not in the base squeak distribution. >> >> >> Begin forwarded message: >> >>> From: St?phane Ducasse <stephane.ducasse(a)inria.fr> >>> Date: November 5, 2008 12:22:57 PM PST (CA) >>> To: Pharo Development <Pharo-project(a)lists.gforge.inria.fr> >>> Subject: [Pharo-project] ArrayedCollection, Boolean >>> SmallInteger>>ccg: cg generateCoerceToOopFrom: aNode on: aStream >>> Reply-To: Pharo Development <pharo-project(a)lists.gforge.inria.fr> >>> >>> I have the impression that we could remove these methods from the >>> images. >>> It seems that this is related to the VM generation? >>> Does someone know more about them? >>> Stef >>> >>> ccg: cg generateCoerceToOopFrom: aNode on: aStream >>> >>> self instSize > 0 ifTrue: >>> [self error: 'cannot auto-coerce arrays with named instance >>> variables']. >>> cg generateCoerceToObjectFromPtr: aNode on: aStream >>> >

1 0

Nov. 6, 2008

Hi 10130 - I cleaned a lot of etoys related things based on pavel submissions. Now it will take long since a lot of packages got published but empty because of a bug in the process I used. Sorry. 10131 313: FileList2 crashes if some subdirectory present Stef

1 0

[Pharo-project] latest update #10129 breaks FileList2 (NonBooleanReceiver: proceed for truth).
by John M McIntosh Nov. 6, 2008

Nov. 6, 2008

After updating my test image and then attempting to open a file list browser I get a walkback This seems to be related to a change in FileList2>>initialize where showDirsInFileList _ false. was dropped by lr 10/29/2008 21:28 Unfortunately showDirsInFileList is not initialized anywhere else so the block then which uses the variable evaluates to nil... -- = = = ======================================================================== John M. McIntosh <johnmci(a)smalltalkconsulting.com> Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com = = = ========================================================================

3 3