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

[Pharo-project] An important infrastructural question
by StÃ©phane Ducasse Sept. 11, 2009

Sept. 11, 2009

HI guys With simon we restarted to code on the new package to substitute packageInfo. Now we have a nice package class which works well (one little bug to fix) and a lot of cool tests. I believe that this was the easy part now we will have to think how to have a migration path and about the possible options. We want your feedback on the following points. Currently PackageOrganizer is not really used or half way and the responsibility are split with packageinfo ending in the kind of mess. We suggest to either use it as a factory and package registry or to remove it. The package manager class (either packageOrganizer or Package class side) should really be responsible to create, query, manage package and emit event. Since packageOrganizer is not really used may be this is easier to move its behavior to PackageInfo (and later Package) class side. events? we think that PackageOrganizer/Package should emit events so that other tools can register. Now the question is how this interact with systemNotification mutating from PackageInfo we could write a script that convert all package info (BTW full of crap) into our package: the question is then what to do with the code that we will load later and relied on * - idea: we keep the * notation but we hook into the tools to create real package. In particular the user could either explicitly add a method extension to a package or use the* notation. The end result should be the star notation in the cat and the use of a real package in the background. - fileout (MC). Similarly we could get rid of the convention at the cat level when fileouting a package. It would work at the package level but fileouting a class would break since the markers would be lost. Also porting code to Squeak or other MC based platform could be a problem. So lukas do you if VW uses the * convention, I imagine at least when saving code? So let us know what you think. Stef

3 4

[Pharo-project] Pavel? ChangeSorter vs ChangesOrganizer
by StÃ©phane Ducasse Sept. 11, 2009

Sept. 11, 2009

Hi Pavel I have the impression that MCPackageLoader>>useChangeSetNamed: baseName during: aBlock "Use the named change set, or create one with the given name." | changeHolder oldChanges newChanges | changeHolder := (ChangeSet respondsTo: #newChanges:) ifTrue: [ChangeSet] ifFalse: [Smalltalk]. oldChanges := (ChangeSet respondsTo: #current) ifTrue: [ChangeSet current] ifFalse: [Smalltalk changes]. newChanges := (ChangeSorter changeSetNamed: baseName) ifNil: [ ChangeSet new name: baseName ]. changeHolder newChanges: newChanges. [aBlock value] ensure: [changeHolder newChanges: oldChanges]. should use Changes MCPackageLoader>>useChangeSetNamed: baseName during: aBlock "Use the named change set, or create one with the given name." | changeHolder oldChanges newChanges | changeHolder := (ChangeSet respondsTo: #newChanges:) ifTrue: [ChangeSet] ifFalse: [Smalltalk]. oldChanges := (ChangeSet respondsTo: #current) ifTrue: [ChangeSet current] ifFalse: [Smalltalk changes]. newChanges := (ChangesOrganizer changeSetNamed: baseName) ifNil: [ ChangeSet new name: baseName ]. changeHolder newChanges: newChanges. [aBlock value] ensure: [changeHolder newChanges: oldChanges]. It would be coherent not to really on UI (changeSorter) but on the domain objects (changeOrganizer). http://bugs.squeak.org/view.php?id= 4825 Stef

1 0

[Pharo-project] BrowserEnvironment is broken?
by Oscar Nierstrasz Sept. 11, 2009

Sept. 11, 2009

Hi Folks, In the latest pharo-web image (pharo1.0-10440-BETAweb09.09.1), the BrowserEnvironment is acting strangely: (BrowserEnvironment new forClasses: (Boolean withAllSubclasses)) selectMethods: [:method | method sendsToSuper ]; label: 'super sends'; open. This lists a whole bunch of methods that do *not* send super, including Boolean>>#&. If I check: (Boolean>>#&) sendsToSuper --> false So what is wrong? This worked ok back at the beginning of July ... - on

3 3

[Pharo-project] [update] #10446
by StÃ©phane Ducasse Sept. 11, 2009

Sept. 11, 2009

10446 ----- ar 315 UIManagerization. Replaces all the trivial references to PopUpMenu, SelectionMenu, CustomMenu, and FillInTheBlank. ar 314 Add a revert item to the patches menu so that local changes can be reverted after diffing to a repository. Useful to weed out intermediate changes that turn out to have no impact after all. bf 316 - sort class vars and pools (merging Eliot s and my fixes) bf 317 - deal with nil user/password in MCHttpRepository - bf 318 - store and read preambles/postscripts in textual source, too - show preambles/postscripts when browsing a snapshot rkrk 319 Making shortcuts work in the MCPatchBrowser tfel.320 Adapt a test-case to reality - In the MCSnapshotBrowser, when we have abolutely nothing selected, we don t want to see an empty string, instead we see the packages scripts or, in the testcase, the info that this package doesn t define any scripts

1 0

[Pharo-project] Rio: binary mode copy not keep
by Vicnet Sept. 11, 2009

Sept. 11, 2009

Hello, I try to use Directory addTree: The src directory I use containts jpeg file. addTree fails because FileKernel use a FileJpeg in InputStream and there is a input pb on this file. I try to add binary mode : base binary addTree: dest but it is the same result: input error. When I copy this file directly with binary mode, it is ok. addTree collect files/dirs to copy but dot not transmit the binary flag. Is it normal ? Directory home binary files first isBinary ==> false ! Should it be true ? Is there another, 'official', way to binary recursive copy ? I use Pharo on Windows plateform. a+ Vicnet -- View this message in context: http://n2.nabble.com/Rio-binary-mode-copy-not-keep-tp3571512p3571512.html Sent from the Pharo Smalltalk mailing list archive at Nabble.com.

5 9

[Pharo-project] Pharo and SFC
by Torsten Bergmann Sept. 11, 2009

Sept. 11, 2009

On goal of Squeak was/still is to become part of the Software Freedom Center. Dont know about the benefits but will/should Pharo work towards that goal too? Thx T. -- Jetzt kostenlos herunterladen: Internet Explorer 8 und Mozilla Firefox 3 - sicherer, schneller und einfacher! http://portal.gmx.net/de/go/atbrowser

1 0

[Pharo-project] little tools support to better browse the squeaktrunk
by StÃ©phane Ducasse Sept. 11, 2009

Sept. 11, 2009

Hi all to push the closure fixes I needed to browse the squeak trunk so I added ScriptLoader new addExtraRepositories so that all the packages get their 3.9a histories and that we can in one click browse the squeak repository. As a test I will integrate some fixes done on the MC browser and other MC related features. I will document that of course in our nice bugtracker. Stef

2 2

[Pharo-project] ODBC: table names and column names/types?
by Schwab,Wilhelm K Sept. 10, 2009

Sept. 10, 2009

Hello all, Is there anything available for Pharo that gives schema information about a database? Bill

4 10

[Pharo-project] namespace collisions
by Chris Muller Sept. 10, 2009

Sept. 10, 2009

Does Pharo have namespaces? If not, may I request that we remember to apply due consideration when choosing names for globals, including class-variables? The top version of ScriptLoader in the Pharo repository consumes the name #Repository, preventing any class or global from using that name. I am now faced with a namespace collision with another package that wants to declare a class named Repository. A class-var in "ScriptLoader" really doesn't need to consume such a valuable name does it? Can this be renamed to something more specific to free up that name? - Chris

2 4

Sept. 10, 2009

Hi even if we integrated the fixes of eliot to fix the ordering class var by MC I still see this behavior: from time to time I get more packages touched just because the order of classVar changes. And now this is a problem when trying to integrate the last fixes issued for the block closure. So I would really appreciate if someone would give a try to fix that. I just have the time to integrate some pending fixes. Stef

2 3