November 2017 - Pharo-dev - lists.pharo.org

[Pharo 7.0-dev] Build #252: 20626-Fixing-Tests-in-CI
by ci-pharo-ci-jenkins2＠inria.fr Nov. 3, 2017

Nov. 3, 2017

There is a new Pharo build available! The status of the build #252 was: SUCCESS. The Pull Request #425 was integrated: "20626-Fixing-Tests-in-CI" Pull request url: https://github.com/pharo-project/pharo/pull/425 Issue Url: https://pharo.fogbugz.com/f/cases/20626 Build Url: https://ci.inria.fr/pharo-ci-jenkins2/job/Test%20pending%20pull%20request%2…

1 0

Baseline/Configuration loading bars
by Jan BlizniÄenko Nov. 2, 2017

Nov. 2, 2017

Hello I have never paid any attention to following thing, but now I use Pharo to teach students object oriented programming and there is one thing that I realized how very bad is from user experience perspective: When I am loading package with lots of dependencies, like Moose or Seaside, it may take multiple minutes (especially on Windows) and those loading bars are there half of a time, but half of a time there is nothing, just nothing - it looks like finished, but it is in fact still loading and everything is therefore sluggish. If I was not sure it is in actually still loading just fine, I would try to kill the process and try again. Example which causes such loading without much of UI is this one: Gofer it smalltalkhubUser: 'zeroflag' project: 'Teapot'; configuration; loadStable. Metacello new repository: 'http://www.smalltalkhub.com/mc/Seaside/MetacelloConfigurations/main'; configuration: 'Seaside3'; version: #'release3.2'; load: #('OneClick'). It would be nice to have some loading bar or something like "loading packages, please wait..." there at all times until baseline/configuration loading is completely done... also some confirming message like "ConfigurationOfSeaside3 loaded successfully" would be nice -- Sent from: http://forum.world.st/Pharo-Smalltalk-Developers-f1294837.html

6 8

UFFI #rebuildFieldAccessors problem
by Aliaksei Syrel Nov. 2, 2017

Nov. 2, 2017

Hi, I have the following line in BaselineOfBloc: (postLoadBloc:) FFIExternalStructure withAllSubclassesDo: #rebuildFieldAccessors. If I don't rebuild accessors, some offsets are nil in external structures. However, on Windows 32bit it fails with the following exception: [image: Inline images 1] How to reproduce: 1. Take Pharo 7 (http://files.pharo.org/get-files/70/pharo.zip) 2. VM (http://files.pharo.org/get-files/70/pharo-win-stable.zip) 3. Excecute FFITestArrayStructure rebuildFieldAccessors Links to Image and VM I took from SmalltalkCI. The reason I'm posting this is because Bloc Appveyour builds fail and more importantly Stef will give lectures in Prague and they will use Bloc for the semester projects. Not being able to install Bloc on Windows is not nice :( We wanted to release alpha version of Bloc and now we can not. Cheers, Alex

2 2

OSWindow OS X VM bug
by Ronie Salgado Nov. 2, 2017

Nov. 2, 2017

Hi, I finally managed to track the problem with keyboard events with OSWindow. However, I cannot fix it without introducing a crash in an autorelease, or a memory leak. The SDL2 problem can be fixed by applying the following patch: diff --git a/platforms/iOS/vm/Common/Classes/sqSqueakEventsAPI.m b/platforms/iOS/vm/Common/Classes/sqSqueakEventsAPI.m index 94c3c4c82..4b5aec8c5 100644 --- a/platforms/iOS/vm/Common/Classes/sqSqueakEventsAPI.m +++ b/platforms/iOS/vm/Common/Classes/sqSqueakEventsAPI.m @@ -100,9 +100,7 @@ sqInt ioProcessEvents(void) { // We need to run the vmIOProcessEvents even if we are using SDL2, // otherwise we end with some double free errors in an autorelease pool. - vmIOProcessEvents(); - if(ioProcessEventsHandler && ioProcessEventsHandler != vmIOProcessEvents) - ioProcessEventsHandler(); + ioProcessEventsHandler(); However, this introduces a crash in an Objective-C autorelease pool, in at leas two locations: diff --git a/platforms/iOS/vm/OSX/sqSqueakOSXApplication+events.m b/platforms/iOS/vm/OSX/sqSqueakOSXApplication+events.m index 4f2ac06ea..c1ec3e1ea 100644 --- a/platforms/iOS/vm/OSX/sqSqueakOSXApplication+events.m +++ b/platforms/iOS/vm/OSX/sqSqueakOSXApplication+events.m @@ -152,9 +152,10 @@ - (void) recordCharEvent:(NSString *) unicodeString fromView: (NSView <sqSqueakO } } - NSString *lookupString = AUTORELEASEOBJ([[NSString alloc] initWithCharacters: &unicode length: 1]); + NSString *lookupString = [[NSString alloc] initWithCharacters: &unicode length: 1]; [lookupString getBytes: &macRomanCharacter maxLength: 1 usedLength: NULL encoding: NSMacOSRomanStringEncoding options: 0 range: picker remainingRange: NULL]; + [lookupString release]; evt.pressCode = EventKeyDown; unsigned short keyCodeRemembered = evt.charCode; diff --git a/platforms/iOS/vm/OSX/sqSqueakOSXOpenGLView.m b/platforms/iOS/vm/OSX/sqSqueakOSXOpenGLView.m index c441332a3..72aa8c42d 100644 --- a/platforms/iOS/vm/OSX/sqSqueakOSXOpenGLView.m +++ b/platforms/iOS/vm/OSX/sqSqueakOSXOpenGLView.m @@ -747,7 +747,7 @@ - (NSString *) dealWithOpenStepChars: (NSString *) openStep { -(void)keyDown:(NSEvent*)theEvent { - keyBoardStrokeDetails *aKeyBoardStrokeDetails = AUTORELEASEOBJ([[keyBoardStrokeDetails alloc] init]); + keyBoardStrokeDetails *aKeyBoardStrokeDetails = ([[keyBoardStrokeDetails alloc] init]); aKeyBoardStrokeDetails.keyCode = [theEvent keyCode]; aKeyBoardStrokeDetails.modifierFlags = [theEvent modifierFlags]; @@ -757,6 +757,8 @@ -(void)keyDown:(NSEvent*)theEvent { [self interpretKeyEvents: down]; self.lastSeenKeyBoardStrokeDetails = NULL; } + + //[aKeyBoardStrokeDetails release]; } In the last location I had to comment my attempt on doing a manual release because it introduces a crash. OSWindow can be tested in a fresh Pharo image by doing the following in a workspace: FFIExternalStructure allSubclassesDo: #rebuildFieldAccessors. > SDL_Event initialize. > SDL2 findSDL2. > OSWindowWorldMorph new open. I have attached the full diff with my bug fix attempt. I hope that someone that is more familiar with the OS X platform code and Objective-C can remove these autoreleases without introducing crashes. When executed in a debugger, the crash is detected in SDL2 Cocoa_PumpEvents: void > > Cocoa_PumpEvents(_THIS) > > { @autoreleasepool > > { > > /* Update activity every 30 seconds to prevent screensaver */ > > SDL_VideoData *data = (SDL_VideoData *)_this->driverdata; > > if (_this->suspend_screensaver && !data->screensaver_use_iopm) { > > Uint32 now = SDL_GetTicks(); > > if (!data->screensaver_activity || > > SDL_TICKS_PASSED(now, data->screensaver_activity + 30000)) { > > UpdateSystemActivity(UsrActivity); > > data->screensaver_activity = now; > > } > > } > > >> for ( ; ; ) { > > NSEvent *event = [NSApp nextEventMatchingMask:NSAnyEventMask >> untilDate:[NSDate distantPast] inMode:NSDefaultRunLoopMode dequeue:YES ]; > > if ( event == nil ) { > > break; > > } > > >> if (!s_bShouldHandleEventsInSDLApplication) { > > Cocoa_DispatchEvent(event); > > } > > >> // Pass events down to SDLApplication to be handled in sendEvent: > > [NSApp sendEvent:event]; > > } > > }} > > > For tracking these actual calls to #autorelease, I had to set a breakpoint on the #autorelease Objective-C selector, and log the stacktrace to find where it is called in our code. Best regards, Ronie

2 1

[Pharo 7.0-dev] Build #251: 20616-Convert-file-format-to-tonel
by ci-pharo-ci-jenkins2＠inria.fr Nov. 2, 2017

Nov. 2, 2017

There is a new Pharo build available! The status of the build #251 was: SUCCESS. The Pull Request #423 was integrated: "20616-Convert-file-format-to-tonel" Pull request url: https://github.com/pharo-project/pharo/pull/423 Issue Url: https://pharo.fogbugz.com/f/cases/20616 Build Url: https://ci.inria.fr/pharo-ci-jenkins2/job/Test%20pending%20pull%20request%2…

2 1

[Pharo 7.0-dev] Build #250: 20616-Convert-file-format-to-tonel
by ci-pharo-ci-jenkins2＠inria.fr Nov. 2, 2017

Nov. 2, 2017

There is a new Pharo build available! The status of the build #250 was: FAILURE. The Pull Request #423 was integrated: "20616-Convert-file-format-to-tonel" Pull request url: https://github.com/pharo-project/pharo/pull/423 Issue Url: https://pharo.fogbugz.com/f/cases/20616 Build Url: https://ci.inria.fr/pharo-ci-jenkins2/job/Test%20pending%20pull%20request%2…

4 3

Layout for placing widgets
by Stephane Ducasse Nov. 1, 2017

Nov. 1, 2017

Hi I was wondering if there is a layout that we can use to propose alternate placing layout for widgets in spec? Like attached to the left 10pixels rubber band box or fixed size box 10pixels attached to the right. Stef

8 21

potentially useful diagrams
by Stephane Ducasse Nov. 1, 2017

Nov. 1, 2017

Hi during the last sprint we were brainstorming how we can help people to understand the git hub setup. I produced these two diagrams with christophe. Are they useful for you? Stef

3 3

[Pharo 7.0-dev] Build #249: 17299 sleep longer than a second in the delay scheduler
by ci-pharo-ci-jenkins2＠inria.fr Nov. 1, 2017

Nov. 1, 2017

There is a new Pharo build available! The status of the build #249 was: SUCCESS. The Pull Request #197 was integrated: "17299 sleep longer than a second in the delay scheduler" Pull request url: https://github.com/pharo-project/pharo/pull/197 Issue Url: https://pharo.fogbugz.com/f/cases/17299 Build Url: https://ci.inria.fr/pharo-ci-jenkins2/job/Test%20pending%20pull%20request%2…

1 0