May 2009 - Pharo-dev - lists.pharo.org

Re: [Pharo-project] 4.1.1beta2u redraw issue

by Sean Allen

i've been playing with this a bit more and if i open a window or minimize one or close one while waiting for a redraw, then the redraw happens. i've loaded up a ton of images and found: last good pharo dev image: pharo0.1-10279dev09.04.3 first bad pharo dev image: pharo0.1-10284dev09.04.4 On May 26, 2009, at 10:12 PM, John M McIntosh wrote: > I'm curious if anyone else has seen this. > > There is some specialized code in the VM, years old to deal with > issues around resizing windows. > > However in the past there has been bugs where Morphic gets confused > about the drawing rules > then *every* pixel draw to the Display is flushed to the VM. That > generates *millions* of draw requests. > > To work around that we (for years) actually flush the drawing surface > to the screen controller process > every N milliseconds so drawing is *slow*, but not like 2 minutes. > Seconds perhaps. > > > On 26-May-09, at 5:09 AM, Sean Allen wrote: > >> >> On May 26, 2009, at 3:57 AM, Damien Cassou wrote: >> >>> On Mon, May 25, 2009 at 6:08 PM, Sean Allen <sean(a)monkeysnatchbanana.com >>>> wrote: >>>> If I resize a window with the Mac 4.1.1beta2u vm, it takes a really >>>> long time to redraw, sometimes up to 2 minutes or so in time. >>>> >>>> horizontally, it does it immediately. this only impact on a resize >>>> that adds/removes space vertically. >>> >>> Have you tried with a previous VM? Have you tried with a previous >>> image? >>> >> yes but not in a scientific fashion. >> >> my previous pharo stuff was rather dated by this point. >> >> where can i download more recent images and vms? i could >> probably pinpoint when it became an issue. >> >> >> _______________________________________________ >> Pharo-project mailing list >> Pharo-project(a)lists.gforge.inria.fr >> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project > > -- > = > = > = > = > = > ====================================================================== > John M. McIntosh <johnmci(a)smalltalkconsulting.com> Twitter: > squeaker68882 > Corporate Smalltalk Consulting Ltd. http:// > www.smalltalkconsulting.com > = > = > = > = > = > ====================================================================== > > > > > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

May 27, 2009

Re: [Pharo-project] 4.1.1beta2u redraw issue

by John M McIntosh

I'm curious if anyone else has seen this. There is some specialized code in the VM, years old to deal with issues around resizing windows. However in the past there has been bugs where Morphic gets confused about the drawing rules then *every* pixel draw to the Display is flushed to the VM. That generates *millions* of draw requests. To work around that we (for years) actually flush the drawing surface to the screen controller process every N milliseconds so drawing is *slow*, but not like 2 minutes. Seconds perhaps. On 26-May-09, at 5:09 AM, Sean Allen wrote: > > On May 26, 2009, at 3:57 AM, Damien Cassou wrote: > >> On Mon, May 25, 2009 at 6:08 PM, Sean Allen <sean(a)monkeysnatchbanana.com >>> wrote: >>> If I resize a window with the Mac 4.1.1beta2u vm, it takes a really >>> long time to redraw, sometimes up to 2 minutes or so in time. >>> >>> horizontally, it does it immediately. this only impact on a resize >>> that adds/removes space vertically. >> >> Have you tried with a previous VM? Have you tried with a previous >> image? >> > yes but not in a scientific fashion. > > my previous pharo stuff was rather dated by this point. > > where can i download more recent images and vms? i could > probably pinpoint when it became an issue. > > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project -- = = = ======================================================================== John M. McIntosh <johnmci(a)smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com = = = ========================================================================

May 27, 2009

[Pharo-project] [ANN] Relicensing Part III

by Gabriel Cotelli

I upload a new slice to Pharo Inbox (includes previous slice): * *Name: SLICE-RelicensingPartIII-GabrielOmarCotelli.2 Author: GabrielOmarCotelli Time: 26 May 2009, 10:22:26 pm UUID: 6623e03e-5542-024a-a380-765fee9a4b29 Ancestors: SLICE-RelicensingPartIII-GabrielOmarCotelli.1 Dependencies: Collections-Strings-GabrielOmarCotelli.20, CollectionsTests-GabrielOmarCotelli.351, Kernel-GabrielOmarCotelli.renggli.333, KernelTests-GabrielOmarCotelli.107* * Additional changes: - Changed raisedToInteger: as suggested by Nicolas. - Re-implemented Number>>integerPart and Number>>fractionPart - Added test cases Please review. Gabriel

May 27, 2009

[Pharo-project] Error: MessageNotUnderstood: ObjectFinalizer>>handle

by Mariano Martinez Peck

Hi folks: I have some objects where I do something link this: WeakRegistry default add: self. In this case, self is an instance of DBXConnection and this class has the handle method. Sometimes, when GC is run, I get this problem: Error: MessageNotUnderstood: ObjectFinalizer>>handle I attach walkback. Is this a bug or I am doing something wrong? Thanks for the help. Mariano

May 27, 2009

Re: [Pharo-project] Crypt ... Libraries ...

by CÃ©drick BÃ©ler

You raise good points but why don't you send this mail to squeak-dev mailing list ? It seems to me that' would be a better audience because you agree on "squeak goals / vision", and IMO this is precisely why people here forked squeak... I think Pharo is not Squeak II at all... but simply an open-source smalltalk that you can work and learn with (in the academic sense)... 2009/5/27 stepken <stepken(a)web.de>: > Hi! > > To make things clearer: > > License Problem: There is a community project with many former contributors. > They all gave away their code for a idea of a free to use smalltalk. > This now is renamed to Pharo by one of those contributors. Now some of those > contributors offer support for that platform. > > If i were one of the former contributors. Who should i accuse for license > infringement? Stef as one of the renamers? No. All other contributors? No. > > I have no idea, what person i could accuse. Who ever will be accused - Every > judge will confirm, that nobody is responsable. Neither for renaming the > project, nor for having contributed code. > > This is the problem for many companies, who own software patents. Free > Software ... there is nobody, they can claim for patent infringement. Linux > - who could be accused? Linus? No. Redhat? No. Novell. No. IBM? No. In what > country? What court is in charge, in what country, what town? Anybody knows. > > So the free software community gives a shit on software patents, rights > infringement. IBM is laughing at all hostile lawyers. They can prove in > front of court, that all ideas in Linux have already been there in one of > their former operating systems. Nothing new. And Linux becomes more and more > popular. > > BSD, GNU, License...who really cares? Richard, yes. But not in case of > Pharo. All open source. So, theoretically seen, question of license is > important, practically not. > > You want to offer commercial software pakets with NDA? Closed Source? With > RUNTIME-Licenses? Oh, yes. Very successful business models of the past. Not > for the future. > > But - i can tell you, who will be suffering. Cincom. They are will be under > pressure by Pharo. You will have to scramble the image, like many commercial > Smalltalks do, to prevent disassembly. > > One of the great advantages of Squeak is, you can find code for everything > in there. Even for sending SMS over Nokia mobiles. Ok, far from being > perfect. But it worked once upon a time. That code you removed ... bad idea! > > >From Pharo I expect: > > Well defined compatibility to other smalltalks. Porting seaside, > databases,... networking code from other smalltalk dialects should be > comfortable. > A well defined and documented GUI, Tutorials, showing, why Morphic is far > superior to MVC, MVP. > Well defined interfaces in Pharo. Ok, what's a interface in a neural network > of sending objects? > Internationalisation. Chinese, Japanese, German, Spanish, Portuguese, French > ... Where are the 'language files' for hobbyists to begin with translation? > Block closures, Traits. > Who cares about a Jitter, wenn it does not run on all platforms. Use LLVM! > Times have changed! Think of china, asia as market. They have godson > processors. (MIPS), no Intel, never!!!! > Working database interfaces for mysql, postgresql, db4o, oracle, db2, goods, > magma ... Tutorials, like that from HPI on seaside. Video tutorials, like > Dolphin has. > Stability. (no hanging GUI when i click on the right upper button to change > window colour). > Renaming back to Squeak or Squeak II. The name 'Squeak' is better known, > already introduced in the world. For Pharo you would have to spend money and > much efforts for advertising. Wrong idea. Why? Go on reading. > ONE Etoys with a nice look. Look at OLPC Etoys. Looks horrible. > Reunification with Scratch. > Tutorials for GUI Programming for Smalltalk beginners on Squeak II > Bringing printing of Morphic canvas, fonts, text to work. > Bringing that mass of lost packages back again into Squeak. Build up a > system of "maintainers", like Debian has. > Well organized 'request for help' database. Precise job offerings. > Percentage metering. > I want to know exactly, what i could contribute. Wheres is the roadmap, the > plan i could understand as 'newby'? Programming, hacking around in smalltalk > is not the problem, i need the idea of the ongoing restructuring process, > the defined interfaces, e.t.c. > Involve HPI, MIT, HP, IBM, APPLE, GOOGLE, NASA ... former contributors, make > them invest into Squeak, into the future of our children. Squeak was and is > an very important international educational project. You can collect > millions of $ for international educational purposes for our children > worldwide, but no money for aÂ additional commercial smalltalk. > Bring those OLPC people to mention the name Squeak in the sense of a great > educational community project in their project. > Learn fundraising. Collect money, much money to drive that development > process. Ever heard of a PAYPAL Button on homepages? Affiliate sytems, Link > exchange? Official Sponsorship. Patronage. Sponsors want to know, where > their money is going to, into what projects within Squeak II. > > And a BIG BANG. > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project > -- CÃ©drick

May 26, 2009

Re: [Pharo-project] Crypt ... Libraries ...

by stepken

May 26, 2009

Re: [Pharo-project] Dictionary use in SmartRefStream

by StÃ©phane Ducasse

I imagine that people did not pay attention. I think that this is important that we polish even the "details". Today I did a lecture and students got excited by Smalltalk the concept and I would love that they get excited by Smalltalk the implementation. This is why beauty in the details is important. So if you have a fix and some tests :) Stef On May 26, 2009, at 11:18 PM, Stephan Eggermont wrote: > While taking a look at Seaside issue 262 I noticed that > > SmartRefStream>>catalogValues: instVarList size: varsOnDisk > > creates a Dictionary with the default size, while having a pretty > good idea of how large it should at least be. Is this called so > little that that it doesn't matter?(it is only called in > SmartRefStream>>readInsanceSize:clsname:refPosn:) > > Stephan Eggermont > > > > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project >

May 26, 2009

[Pharo-project] Dictionary use in SmartRefStream

by Stephan Eggermont

While taking a look at Seaside issue 262 I noticed that SmartRefStream>>catalogValues: instVarList size: varsOnDisk creates a Dictionary with the default size, while having a pretty good idea of how large it should at least be. Is this called so little that that it doesn't matter?(it is only called in SmartRefStream>>readInsanceSize:clsname:refPosn:) Stephan Eggermont

May 26, 2009

Re: [Pharo-project] Crypt ... Libraries ...

by John M McIntosh

On 25-May-09, at 4:04 PM, stepken wrote: > https, md5, DES, AES .... i really wonder, why you can't decide to > plugin libcrypt ....? > > In C programming language we have all, we can imagine. So why reinvent > the wheel by rewriting complex code in smalltalk? The problem with providing plugins is that it's very non-portable. Issues: (a) The number of people who write and maintain plugins for Squeak is likely countable on a pair of hands. (b) For Sophie we had to statically link FreeType into Sophie because Apple over different operating systems versions provided a multiple selection of broken freetype versions. Witness the hassles people have installing a product on linux because of incompatible libraries. (c) Does that plugin work with os-x 10.2 on a powerpc machine? Gee why not? Who knows? Where is that powerPC 10.2 test machine? Obviously if the code is all in smalltalk the above is not a problem, and you can fix it, If all in FFI then less of a problem and maybe you still can fix it. If a plugin, then I doubt you'll fix it. Certainly we don't expect to take *everything* and rewrite in Smalltalk, witness the freetype and curl work as an example, and all of this doesn't touch on license insanity and wondering if that library is *safe* to use. -- = = = ======================================================================== John M. McIntosh <johnmci(a)smalltalkconsulting.com> Twitter: squeaker68882 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com = = = ========================================================================

May 26, 2009

Re: [Pharo-project] Crypt ... Libraries ...

by StÃ©phane Ducasse

Pharo is getting hot. It seems that we have our own little troll. Champaign everybody :) Stef > Sorry folks, can't help myself. Reactions below.... > > On May 26, 2009, at 6:12 AM, stepken wrote: > >>> Pharo developers are few and limited. They are trying to do all of >>> their >>> best on Pharo. We know this an important package but perhaps they >>> have more >>> important things to do. Pharo is not even in a beta of first >>> milestone. This >>> can be planned in a future milestone. >> >> Hmmm. How many manyears were invested in Squeak? 1000, 2000? >> >> You have removed lots of code due to "license problems". See R.A. >> Harmon code. What a pitty to throw away code ... >> >> Do you really think, he really cares about his former code >> contribution to a system, which from the beginning was free and open? > > So anytime you want to, you can just decide what someone else's > intention for their code contribution was and do what you want? Each > person owns his own code, and Pharo has to be careful and have actual > permission to use the code they distribute. > > >> >> Don't you think, every jugde would confirm, that such code >> contributions were made in the sense of "giveaways"? > > Are you going to pay for the lawyers necessary to defend such an > untenable argument? > >> >> I think you don't have a real understanding of what you have >> destroyed now. >> >> Having rewritten array/colletions/streams code ... a bit shorter >> code now, because of traits. Who really cares? > > People write a lot of code because they desire to; what is the point > of Traits if you never use it in your system? > >> >> You want to build a commercial platform to make money. Supported. >> Yes. > > I have paid (and still am) several squeak/pharo developers for support > on these systems. Dolphin is windows only, which kills it for a lot of > people. > >> >> Didn't you all Smalltalk Developers notice, that even Dolphin was >> close to give up? They hat to find their new business model. And >> they also have a very sophisticated *free* version. >> >> GNU Smalltalk 3.0 ... fine thing, free, can run seaside, pier, .... >> what do i need more? Free! > > No, that is GPL, which has plenty of freedom restrictions. > >> >> Do you really hope to get one cent for supporting a free smalltalk >> platform? >> > > Absolutely they will. > >> I really appreciate your effords. There are things, that had to be >> done on Squeak. Traits, 5th implemention of closures ... > > How many implementations of closures? what are you talking about? Do > you take each individual person who had a project on this platform > assume that was the will of the community? > >> >> But ... you haven't included the old code contributors at Viewpoint >> Research Institute, MIT (Scratch), HPI Berlin. > > Pharo has some very specific goals for a clean, well factored, test- > driven base Smalltalk system. I suggest you take that into account; > what you want is a giant unmaintainable mess of spaghetti. > >> >> Where are your plans to give back your code changes to squeak/ >> scratch community? No. Very egoistic, IMHO. > > All the code is free and available, and there is no stopping of > squeak/ > scratch/other community members reintegrating anything they like. > >> >> Where is the "soft refactoring" without destroying anything ...? At >> the moment ... i see some developments with pleasure, others are >> catastrophic ... >> >> Eliot writing the 10th implementation of a squeak jitter. Fine. >> Every profesional company uses LLVM. Apple e.g. > > Again, what are you talking about? I've never heard of 9 previous jit > implementations. In one sentence you complain that Pharo throws away > code and doesn't use stuff from every other squeak based system, and > then you want the whole system re-implemented on LLVM? Wow. > >> >> I wanted to know, how fast a programming language with jitter could >> be done. It took me about 2 weeks ...LL(1) grammar, LLVM ... works >> like a charm on Intel, ARM, PPC ... and really fast! Other platforms >> to be tested. Of cause, lots of things to be done .. debugging ... >> profiling ... no classbrowser ... but i can run older squeak code >> with morphic. 80%/20% problem ;-) Or rather 90/10? Dunnow, i stopped >> the development. Was an experiment. >> > > Interesting... where is the code? is it free? why didn't you finish > it? Maybe Eliot or someone else could benefit from it, but not now. > >> Will Eliots jitter run on different processor architectures? No! 386 >> machine code only. So - no real portability! Who really cares that >> stuff then? Ever heard of the giant chinese market? They use MIPS >> clone 32+64Bit processors with Linux now. Everywhere. China can't >> officially use Intel, because of too much power consumption. I've >> been there several times. MIPS, everywhere. In China and Japan you >> find lots of people doing squeak stuff. Potential code contributors >> code developers for Pharo, partners! > > It all comes back to time and money. Eliot is being paid by Qwaq, and > they are generously allowing all that work to go back to the > community. Who are you funding to do work? Or what work are you > contributing? > >> >> Morphic code .. from SELF Programming Language development ... then >> ported to squeak, then to Javascript. Fine. ETOYS on TOP. >> >> Where is my loved ETOYS? The only reason to use Squeak or Pharo! >> Education was and is the domain of Squeak. Nothing else. > > Squeakland.org - > OLPC Project (ever hear of it?) > >> >> I asked Frank Lesser for EToys / Morphic on top of his Smalltalk. >> Matured, blindingly fast, Jitter written in Assembler. Now the base >> for DNG. Not a great problem, he meant, he had it once ported for >> testing purposes. >> >> If i really needed a sophisticated smalltalk to start new >> projects ... what do you think, what i would take? Smalltalk for >> educational purposes, e.g.? I have free choice. But no ETOYS, >> anywhere. Children really love EToys. It's a marvelous software >> package for educational purposes. >> >> Many companies are caught in their own jail of huge masses of GUI >> code. They can't go with other smalltalks. Will they ever change to >> a "supported Pharo"? I think: No. >> >> Morphic ... no MVC, no MVP. What has happened to Tweak code? ETOYS >> was ported to that GUI. That code was quite ok. Why haven't the code >> autors spent that code for Squeak/Pharo? > > First you have to be able to remove Morphic from it's death grip on > the system and have packages all the way down. Then start integrating > new GUI frameworks. > >> >> So please ... don't tell me you haven't enough hands at the moment. >> Do the right things, not just some few things right, beautifying >> GUI, implemeting the 5th version of closures and well ... traits. > > It sounds like you need to start you own Smalltalk system, based on > LLVM, or your own super fast jitter that you wrote in two weeks. Then > you can crowdsource all you want. Lot's of people are participating > and things are going swimmingly. > > > The tone of your emails are offensive and you cast aspersion on > everyone working on the Pharo project. If you don't like it, don't > read the list and don't use it. > >> >> Do 'crowdsourceing', invite people to contribute, learn community >> organizing first. Give them a reason to contribute. Give them a >> future, a business modell to participate. >> >> Your egoistic path to make a commercial supported free smalltalk ... >> good idea, but you are lightyears far from that. >> >> Leo Penta wrote a nice book about that 'crowdsourceing' stuff, or >> reread "Eric Raymond - Cathedral and the basaar" again and again and >> again. *PARTICIPATION* Bring people to participate your business >> model. >> >> LEARN from the mistakes, many, many good programmers did. See TWEAK >> bad fate. Dead development process. Good sourcecode, lost in >> space ... like many squeak packages. >> >> *Defective development process*. I posted that as issue, it was >> deleted by someone not really understanding, what i meant. >> >> thanx for understanding. >> >> cheers, Guido Stepken >> >> >> _______________________________________________ >> Pharo-project mailing list >> Pharo-project(a)lists.gforge.inria.fr >> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project > > > _______________________________________________ > Pharo-project mailing list > Pharo-project(a)lists.gforge.inria.fr > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project >

May 26, 2009