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

[Pharo-project] should the PharoDev hudson image be buld using the baseline instead?
by Mariano Martinez Peck Jan. 18, 2011

Jan. 18, 2011

Hi. A trade-off here. Right now, the PharoDev image from Hudson is being built using ConfigurationOfPharo project lastestVersion. This is more or less good because we have the last version of the ConfigurationOfPharo. But at the same time, updating the versions of ConfigurationOfPharo is tedious. So, what about loading the baseline istead? So that metacello automatically load the latest version of each package/project ? The good thing is that every change in external configurations/packages will be automatically included. But at the same time this is a disavtange because we can build crap (someone commits something that doesn't work). When updating ConfigurationOfPharo by hand we usually test it before.... anyway, isn't this the idea if hudson? if it builds crap, we want to know it, as soon as possible. NOBODY should use a Hudson image (as it is now) for a stable Pharo1.2 image. what do you think? cheers mariano

3 2

[Pharo-project] Issue 3517 in pharo: Failing test AutomaticMethodCategorizerOB
by pharo＠googlecode.com Jan. 18, 2011

Jan. 18, 2011

Status: Accepted Owner: marcus.denker Labels: Milestone-1.2-DevImage New issue 3517 by marcus.denker: Failing test AutomaticMethodCategorizerOB http://code.google.com/p/pharo/issues/detail?id=3517 >>> AutomaticMethodCategorizerOB.Tests.OBCmdAutoCategorizeWithAMCTest.testOBAutocategorizeCommandOverride >>> >>> 0.014 15

1 1

[Pharo-project] Issue 3547 in pharo: Have a look at the new match: implementation
by pharo＠googlecode.com Jan. 18, 2011

Jan. 18, 2011

Status: Accepted Owner: stephane...(a)gmail.com Labels: Milestone-1.3 New issue 3547 by stephane...(a)gmail.com: Have a look at the new match: implementation http://code.google.com/p/pharo/issues/detail?id=3547 Collections-ul.420 Hi Squeakers, Latest trunk version broke the Aida/web and Herbert managed to find the reason in new behavior of String>>match: which doesn't allow two '**' in pattern anymore. In Aida the matching patterns are usually composed out of many strings and as it happens two stars occurs in Aida's default installation already. Because this new match behavior (not allowing two '**') is new only in Squeak but allowed in all other Smalltalks, I propose to reconsider this decision and better retract back to old behavior. answer: ------ The old behavior was definitely wrong. Here are two examples with the old behavior from the corresponding issue (http://bugs.squeak.org/view.php?id=6841 ): '*#' match: 'e'. "false" '**' match: 'e'. "false" If you look at the previous version of the method from 2004 (I don't know if this is the original or not), then you'll see that it tried to raise an error if the string contained '**', but it failed to do so because of this bug. So the intended behavior - to disallow ** in the pattern - is not new at all. IMHO the best solution to this problem is to rewrite the method to allow any combination of * and # in the pattern.

1 3

Re: [Pharo-project] Xtreams-Xtras: hashing and HMAC
by mkobetic＠gmail.com Jan. 18, 2011

Jan. 18, 2011

"Sven Van Caekenberghe"<sven(a)beta9.be> wrote: > > I didn't find a way to browse a "package" in Pharo 1.2 so I'm struggling a bit to keep the categories in line. > > Hmm, click the 'Browse' button in either the Monticello Browser or the Repository Browser with the correct package selected ? But I guess you know that as well... Well, those seem to be "read-only", i.e. any edits I try to make in those seem to disappear, so I've been ignoring those browsers. Interestingly, the one-click image (Pharo 1.1.1) seems to have 'browse package' in the SystemBrowser menus which seems to be more functional, but that doesn't seem present in the PharoCore 1.2, so I'm just getting by with the basic SystemBrowser atm. > The problem is the one I reported earlier: XTFileReadingWritingTest expecting a read stream to be aware of data written to a file after it was opened. This does not seem to work on Mac OS X. Yeah I have your previous report still in my inbox, but I'm kinda stuck there. As I said I don't have OSX to work with so I can't dig in to figure out what's going on. Given that it works on OSX in VW, I'm inclined to say that it should work on OSX in Squeak too. I wonder if there's some mixup in terms of file descriptors, maybe we're not talking to the same one or something such. I think Nicolas has OSX available so I'm hoping he might be able to find the time to look into it at some point.

2 1

[Pharo-project] Surprising binary messages!
by Noury Bouraqadi Jan. 18, 2011

Jan. 18, 2011

5 is the surprising answer of the expression: 6 + 4 / 2 It should be 8 But, I got 5 in both 1.1 and 1.2. Noury

7 10

Re: [Pharo-project] [squeak-dev] Re: Xtreams-Xtras: hashing and HMAC
by mkobetic＠gmail.com Jan. 18, 2011

Jan. 18, 2011

"Sven Van Caekenberghe"<sven(a)beta9.be> wrote: > > The code is in Xtreams-Xtras and Xtreams-XtrasTests (note that you'll have to load FFI!). I tested it on Pharo 1.2 beta with both cog vm and standard vm. There are few test failures with file terminals that don't seem to like to be closed twice. Should be an easy fix. > > Could it be that you forgot to check in #hashing: for read streams ? Or did I load the code while you were updating ? Ah, my bad, I misspelled the category on XTReadStream and so it was omitted. I didn't find a way to browse a "package" in Pharo 1.2 so I'm struggling a bit to keep the categories in line. Thanks for the note, a new version of Xtras is on squeaksource.

2 1

[Pharo-project] Metacello Help book in Pharo-Dev
by Torsten Bergmann Jan. 18, 2011

Jan. 18, 2011

>the current behavior of ConfigurationOfPharo loads Metacello >completly, included packages like Metacello-Help and Metacello-ProfStef OK, then its fine >someone can update that to 1.0-beta.28 and you will see the new versions. When I switch to Metacello "1.0-beta.28" in ConfigurationOfPharo then it fails while loading OB. A conflict since Metacello uses OB too, maybe in another version? Thx T. -- NEU: FreePhone - kostenlos mobil telefonieren und surfen! Jetzt informieren: http://www.gmx.net/de/go/freephone

3 2

[Pharo-project] [ANN 1.2] pre-build Core 1.2rc1 #12311
by Marcus Denker Jan. 18, 2011

Jan. 18, 2011

https://gforge.inria.fr/frs/download.php/28162/PharoCore-1.2rc1-12311.zip -- Marcus Denker -- http://www.marcusdenker.de INRIA Lille -- Nord Europe. Team RMoD.

2 1

[Pharo-project] Xtreams-Xtras: hashing and HMAC
by mkobetic＠gmail.com Jan. 18, 2011

Jan. 18, 2011

I started playing with FFI and got the hashing streams working. They require OpenSSL's libcrypto, which should be available on any reasonably recent Unix and OSX. The VW version also has interface to native Windows bcrypt.dll (available on Vista and later) but I haven't done that part yet. Anyway the hashing streams seem to work and performance isn't bad either. Hashing 100MB of ones [ (([ 1 ] reading limiting: 100000000) hashing: 'md5') -= 0; close; digest ] timeToRun takes little over 5 seconds on not particularly fast machine. The above expression runs about 2.5 times faster on VW on the same machine, but that's probably due to the time spent in the block based terminal. Profile on VW says that only about 15% of the time is spent in the hashing layer. Hashing the 21M pharo image is pretty much instant: ((FileDirectory default / 'pharo.image') reading hashing: 'md5') -= 0; close; digest The code is in Xtreams-Xtras and Xtreams-XtrasTests (note that you'll have to load FFI!). I tested it on Pharo 1.2 beta with both cog vm and standard vm. There are few test failures with file terminals that don't seem to like to be closed twice. Should be an easy fix. Enjoy, Martin P.S.: If any resident FFI experts find the time and interest to look over my C bindings and find something wrong, please let me know. In particular I have't quite figured out how to work with struct fields that contain arrays (e.g. 'unsigned char [128]'). I figured out that I can declare the actual size of the field as a third parameter in the #fields declaration, but the accessors generated for the field are not right. I suspect they need to be some form of ExternalData but I don't quite grok how that should be created (see e.g. HMAC_CTX>>key). Also it seems that there isn't a type definition for 'int' so I'm replacing all references to that with 'long'. Is that how it's done?

2 1

Re: [Pharo-project] [squeak-dev] more Info in aboutSqueak
by Alexander LazareviÄ Jan. 18, 2011

Jan. 18, 2011

2011/1/13 Enrico Spinielli <enrico.spinielli(a)googlemail.com> > H Alex, > great idea! > > As a side thought, couldn't it be a dialog with cut&past-able text? > I thought about that too, but now used a small ToolBuilder window instead. In the SystemReporter you can chose what system information should be gathered. So you could have a compact overview of the image and the VM or a full length/all details report. Add the repository SystemReporter to Monitcello. MCHttpRepository location: 'http://www.squeaksource.com/SystemReporter' user: '' password: '' Load the package SystemReporter and do SystemReporter open in a workspace. This very simple and needs to be polished further. License is MIT and everybody is invited to add code. It's easy to extend the reporter and it is explained in the class comment. This works with Squeak Trunk and Pharo 1.1.1 (needs to get rid of the deprecation warning) Alex

1 1