March 2012 - Pharo-dev - lists.pharo.org

[Pharo-project] Fuel - class loading issue when superclass changed inst vars
by Martin Dias March 6, 2012

March 6, 2012

Hello, I'd like to discuss about a problem found when loading classes with Fuel (for people who don't know it: http://rmod.lille.inria.fr/web/pier/software/Fuel) Fuel can serialize a class either as a global or as a "regular" object. In the latter case, it saves the class with its compiled methods, including their bytecodes. (This was demonstrated in http://marianopeck.wordpress.com/2011/09/24/importing-and-exporting-package… ) Well, the issue I want to talk about happens when loading a class with a method that uses an inherited instance variable. In a short example: - define class A with instVars 'x' - define class B with a method that uses 'x'. - serialize B in a file - remove B from system - redefine A with instVars 'y x' - materialize B from the file -----> the bytecodes of B's method, that should refer to 'x', instead refer to 'y' (because inst vars are encoded in bytecodes just as indexes) Of course, there are variations of this problem: the variable was whether removed or renamed. Some solutions: 1) detect the change in a superclass and then: a) just raise an error b) compile based on attached source c) decompile parse tree from bytecodes, fix it, and generate the method. 2) serialize the parse tree *instead* of bytecodes, and always generate the method when materializing. I am thinking on the use case of package sharing, like Parcels. Note that since Fuel supports multiple strategies for serializing the same object, it is not necessary to choose just one option to fit in every use case. - I'm not sure, but I guess Parcels makes something like 1c). - The alternative 1a) doesn't require any compiler (in a bootstrap image is good) - I am tempted to explore 2), but not sure if that makes sense. I am curious about your opinions. Best regards, MartÃn

3 2

Re: [Pharo-project] Pharo Kernel break...
by Marcus Denker March 6, 2012

March 6, 2012

On Mar 6, 2012, at 11:08 AM, Pavel Krivanek wrote: > I created a fix > > http://code.google.com/p/pharo/issues/detail?id=5438 > Thanks! I will have a look this evening. -- Marcus Denker -- http://marcusdenker.de

1 0

[Pharo-project] [update 1.4] #14379
by Marcus Denker March 6, 2012

March 6, 2012

14379 ----- Issue 5321: DataStreamTest creates files that are not deleted http://code.google.com/p/pharo/issues/detail?id=5321 Issue 5285: HostWindowTests>>testOne (workaround) http://code.google.com/p/pharo/issues/detail?id=5285 Issue 5153: warningAllowed does not work with undefined selectors http://code.google.com/p/pharo/issues/detail?id=5153 -- Marcus Denker -- http://marcusdenker.de

1 0

Re: [Pharo-project] Pharo Kernel break...
by Marcus Denker March 6, 2012

March 6, 2012

On Mar 4, 2012, at 10:32 AM, Pavel Krivanek wrote: > Hi, > > I shortly looked at it. To reproduce the key problem without need to > start shrinking process we may use this expression. > > Behavior removeCategory: '*Deprecated14' Strange... a workaroung would do to just unload it with Monticello instead. We should merge the shrinking of #cleanUpForProduction with the one of the Kernel... Marcus -- Marcus Denker -- http://marcusdenker.de

2 2

[Pharo-project] Nautilus collisions in icons
by Mariano Martinez Peck March 6, 2012

March 6, 2012

Hi ben. I found that method's icons are not only one, and there can be colisions. For example, I have test methods which came from a trait. So I have the trait icon rather than the test. And I can imagine some other colision as well ... So...what can we do? can we have a list of icons? I mean..have as much icons as requiered? (instead of having colisions and let only one) If we cannot have multiple...then I would let the test icon in the case of traits for example....because I don't care where the method come from...the same way I don't care if it comes from a superclass, I don't care it comes from a trait. But I do care that if it is a test, I could do cmd+t and see the result :) what do you think? Cheers -- Mariano http://marianopeck.wordpress.com

2 1

[Pharo-project] NativeBoost VM's
by Jan van de Sandt March 5, 2012

March 5, 2012

Hello, On the Pharo Jenkins server [1] there is a recent build of a VM with NativeBoost support for the Mac. But the Unix version is six months old and there is no Windows version. Is this just a Jenkins issue and is NativeBoost working on Unix and Windows? Or are these platforms not working yet? Jan. [1] https://ci.lille.inria.fr/pharo/view/NativeBoost/

3 3

Re: [Pharo-project] Showing the setUp method in Nautilus
by Alexandre Bergel March 5, 2012

March 5, 2012

Voila, I produced version 4.1 Please check. I did not use a plugin since the commands are directly implemented on NautilusUI, I therefore guess that a plugin is not the way to define commands. Also, something that I would find very useful, is to open a new Nautilus directly on the Group tab. I often see myself switching to the Group tab. Am I the only one? Cheers, Alexandre On 5 Mar 2012, at 13:40, Benjamin wrote: > > On Mar 5, 2012, at 12:52 PM, Alexandre Bergel wrote: > >> Hi! >> >> I worked on a new method item action to easily show the setUp method of a testCase. This helps reduce the amount of clicks and context switching when one wants to know what the setUp method is about while writing a test method. > > Cool > >> I think it is a good candidate for being included in Nautilus. >> Ben, if you agree with this, I can to commit in the Nautilus rep and produce a new version of the configuration. How does that sound? > > Commit, and I will have a look ;) > I though you started also a NautilusPlugin related to that, am I wrong ? > >> >> Question: how to associate a shortcut to an action? > > Anywhere on class side something like: > > buildRefactoringMethodShortcutsOn: aBuilder > <keymap> > > (aBuilder shortcut: #rename) > category: #NautilusMethodShortcuts > default: $r command > do: [:target || scroll | > scroll := target methodWidget vScrollValue. > target refactor renameMethodFor: target selectedMethod. > target methodWidget vScrollValue: scroll] > > The categories are: > #NautilusMethodShortcuts > #NautilusProtocolShortcuts > #NautilusClassShortcuts > #NautilusPackageShortcuts > > and here target is the NautilusUi instance which get the keyStroke. > >> >> Side comment: it would be a good move to adopt a class hierarchy a la OB to define commands. Defining all the commands in NautilusUI does not scale and simply does not work for non trivial command. > > For me, it generates to much classes. And now that Guillermo did a great work for KeyMappings, I prefer to use it :) > > > Ben > >> >> Cheers, >> Alexandre >> -- >> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: >> Alexandre Bergel http://www.bergel.eu >> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. >> >> >> >> <ShowSetupInNautilus.1.cs> > -- _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: Alexandre Bergel http://www.bergel.eu ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.

4 6

[Pharo-project] MVP
by Schwab,Wilhelm K March 5, 2012

March 5, 2012

Hello all, I have been to busy to read today, but I like seeing any discussion of MVP. I am **slowly** working on a framework for it, but would be glad to be beaten to it. Default models are key; they must be replaceable, but (sub)triads should work "out of the box." Models can include value holders, value adapters, converters, and contexts, depending on how they are wired. I have something of a start on the adapters and converters, which I can try to package and make available. If nothing else, they will have value in seeding discussion if not be directly usable. More to come. Bill

5 21

[Pharo-project] Cog VM binaries as per VMMaker.oscog-eem.152/r2538.
by Eliot Miranda March 5, 2012

March 5, 2012

Cog VM binaries as per VMMaker.oscog-eem.152/r2538<http://www.mirandabanda.org/files/Cog/VM/VM.r2538/> . Fix tricky context state bug that can cause crashes in the GC. Contextpart>runUntilErrorOrReturnFrom: may pop the last argument. Hence if a frame is married in this state its spouse context won't have its last argument slot initialized. If later the frame is divorced and the arguments are not updated the last argument slot can be left with this uninitialized slot and ... bang. The solution is to update arguments as well as stack contents when divorcing. This is good also because it removes the VM's assumption that method arguments are read-only, and that's only enforced by the Smalltalk bytecode compiler, not by the bytecode. -- best, Eliot

3 3

Re: [Pharo-project] MVP
by Marcus Denker March 5, 2012

March 5, 2012

On Mar 5, 2012, at 7:25 PM, Schwab,Wilhelm K wrote: > At least in the 1.4 seaside image, I do not get color highlighting the browser - debugger and workspaces do show > coloring. I just grabbed the latest seaside image to verify. > then the build scripts of that image are wrong. A "seaside" image should not load anything related to browsers, workspaces or debuggers. Marcus -- Marcus Denker -- http://marcusdenker.de

3 2