- Pharo-dev - lists.pharo.org

[Pharo-project] Alt-. does not work anymore on Linux or?

by Torsten Bergmann

>I can try. What I find a bit "odd" is that when I press Alt-. it >actually prints a $. (say in a Workspace) just as if I had *only* >pressed dot. That may be the main problem here - Alt is not properly >detected? Yes same here. But some Transcript outputs into UserInterruptHandler>>handleEvent: and you will notice that it comes with different "(evt at: 4)". Yes - to me it also looks as if ALT is not properly detected ... The default interrupt key value in UserInterruptHandler(class)>>new is ($. asciiValue bitOr: 16r0800) So I guess this bit enabeling is for ALT (CMD). Dont know why we dont use a shared pool with proper constants KEY_CMD_BIT := 16r0800 ($. asciiValue bitOr: KEY_CMD_BIT) which would be much clearer. At least to me. The code in UserInterruptHandler>>handleEvent: could profit from constants as well, I hate these "(evt at: 4)" and "(evt at: 3)" and "(evt at: 5)" with bit manipulation using magic 8 ...

Nov. 29, 2012

Re: [Pharo-project] Getting started with Athens

by Ben Coman

Igor Stasenko wrote: > Make sure you enabled freetype library in settings. > > (LogicalFont familyName: 'Arial' pointSize: 20) realFont. > > should answer a freetype font, not strike font instance. > > Perhaps that can be added somewhere in initialization code as an assert with a useful exception message.

Nov. 29, 2012

Re: [Pharo-project] Alt-. does not work anymore on Linux or?

by David T. Lewis

On Thu, Nov 29, 2012 at 12:28:09PM +0100, G?ran Krampe wrote: > Hey! > > I just realized that alt-. doesn't seem to fly on my Ubuntu 12.10 in > Pharo 1.4 nor 2.0 one-clicks. > > It *does* however work fine in Squeak 4.3 one-click. > > In Squeak I can: > Run "(Delay forSeconds: 5) wait" and interrupt it with alt-. > Or just press Alt-. and get a debugger. > > Also it seems that Process browser in Squeak is opened up *without* > having any process be suspended. Not sure why Pharo is different there > and shows the UI process as suspended? > > regards, G?ran > I do not know if this is relevant to Pharo any more, but there is quite of bit of information about problems of this type in the links starting at http://bugs.squeak.org/view.php?id=7655. Maybe this will give some ideas for debugging. The four test expressions that should be interruptable (per Andreas Raab) are: [true] whileTrue [[true] whileTrue] forkAt: Processor userSchedulingPriority + 1 Smalltalk createStackOverflow [Smalltalk createStackOverflow] forkAt: Processor userSchedulingPriority + 1 Extra credit if anyone can figure out how to turn those expressions into unit tests ;-) HTH, Dave

Nov. 29, 2012

Re: [Pharo-project] Alt-. does not work anymore on Linux or?

by GÃ¶ran Krampe

On 11/29/2012 01:57 PM, GÃ¶ran Krampe wrote: > On 11/29/2012 01:51 PM, Torsten Bergmann wrote: >> Same issue with Stack VM. So it looks like it >> is image side. > > Yes, stackVM did not help, at least not with Pharo 2.0. > >> When the keyboard focus is in a workspace and >> I hit CMD+. it not directly interrupts. Only >> when I repeat very often. >> >> I've put a >> >> Transcript show: toInterrupt asString; cr. >> >> into UserInterruptHandler>>handleUserInterrupt and >> very often this block temporary variable is nil since >> #processToInterrupt returns nil. >> >> Sometimes the #processToInterrupt catches >> a process to interrupt and then the debugger opens. >> >> GÃ¶ran can you confirm that as well? Yes, similar behavior. I also noticed that Pharo (compared to Squeak) does not send #primInterruptSemaphore: ... Not sure if related, but perhaps. regards, GÃ¶ran

Nov. 29, 2012

Re: [Pharo-project] Alt-. does not work anymore on Linux or?

by GÃ¶ran Krampe

On 11/29/2012 01:51 PM, Torsten Bergmann wrote: > Same issue with Stack VM. So it looks like it > is image side. Yes, stackVM did not help, at least not with Pharo 2.0. > When the keyboard focus is in a workspace and > I hit CMD+. it not directly interrupts. Only > when I repeat very often. > > I've put a > > Transcript show: toInterrupt asString; cr. > > into UserInterruptHandler>>handleUserInterrupt and > very often this block temporary variable is nil since > #processToInterrupt returns nil. > > Sometimes the #processToInterrupt catches > a process to interrupt and then the debugger opens. > > GÃ¶ran can you confirm that as well? I can try. What I find a bit "odd" is that when I press Alt-. it actually prints a $. (say in a Workspace) just as if I had *only* pressed dot. That may be the main problem here - Alt is not properly detected? regards, GÃ¶ran

Nov. 29, 2012

[Pharo-project] Alt-. does not work anymore on Linux or?

by Torsten Bergmann

Same issue with Stack VM. So it looks like it is image side. When the keyboard focus is in a workspace and I hit CMD+. it not directly interrupts. Only when I repeat very often. I've put a Transcript show: toInterrupt asString; cr. into UserInterruptHandler>>handleUserInterrupt and very often this block temporary variable is nil since #processToInterrupt returns nil. Sometimes the #processToInterrupt catches a process to interrupt and then the debugger opens. GÃ¶ran can you confirm that as well? Bye T.

Nov. 29, 2012

Re: [Pharo-project] Alt-. does not work anymore on Linux or?

by Camillo Bruni

could you try with the stackVM? http://pharo.gforge.inria.fr/ci/vm/stack/ with cog I have similar issues that often I cannot interrupt, whereas it works with the stack VM... On 2012-11-29, at 09:15, Torsten Bergmann <astares(a)gmx.de> wrote: > On Win32 (german locale) I can not interrupt > > (Delay forSeconds: 10) wait. 1 inspect > > when using NBCog.exe as a VM. > > Even when I set a new key like ";" (without ALT) I can not > interrupt the delay. > > |handler| > handler := UserInterruptHandler new setInterruptKeyValue: ($; asciiValue). > handler registerIn: InputEventFetcher default. > (Delay forSeconds: 10) wait. 1 inspect > > I can not interrupt with ";" while the delay is waiting. But it interrupts > aftewards when I hit ";". > > I'm not sure it is really a locale problem as described in > > http://code.google.com/p/pharo/issues/detail?id=6688 > > When I try 1.4 one click I have the same problem - the delay > is not interruptable. > > When I try 1.3 one click I can interrupt. Can it be related > to switching to Cog VM starting from 1.4? > > Bye > T. >

Nov. 29, 2012

[Pharo-project] Alt-. does not work anymore on Linux or?

by Torsten Bergmann

On Win32 (german locale) I can not interrupt (Delay forSeconds: 10) wait. 1 inspect when using NBCog.exe as a VM. Even when I set a new key like ";" (without ALT) I can not interrupt the delay. |handler| handler := UserInterruptHandler new setInterruptKeyValue: ($; asciiValue). handler registerIn: InputEventFetcher default. (Delay forSeconds: 10) wait. 1 inspect I can not interrupt with ";" while the delay is waiting. But it interrupts aftewards when I hit ";". I'm not sure it is really a locale problem as described in http://code.google.com/p/pharo/issues/detail?id=6688 When I try 1.4 one click I have the same problem - the delay is not interruptable. When I try 1.3 one click I can interrupt. Can it be related to switching to Cog VM starting from 1.4? Bye T.

Nov. 29, 2012

Re: [Pharo-project] Alt-. does not work anymore on Linux or?

by Sebastian Nozzi

Thanks. Now I see why my image hung when doing: true whileTrue: [(Delay forSeconds: 5) wait. do-something-else] I also tried to interrupt it, but was not able. Only in the "do-something-else" it would have worked, but the time-frame was too small. 2012/11/29 Torsten Bergmann <astares(a)gmx.de>: > It's not a Linux issue. When I run > > (Delay forSeconds: 5) wait > > on Win32 with Pharo 2.0 I can also not > interrupt it using ALT+. > > I checked and "UserInterruptHandler cmdDotEnabled" > returns true in my image. Havent digged deeper. > > Can you please open an issue? > > Thx > T. >

Nov. 29, 2012

[Pharo-project] Alt-. does not work anymore on Linux or?

by Torsten Bergmann

It's not a Linux issue. When I run (Delay forSeconds: 5) wait on Win32 with Pharo 2.0 I can also not interrupt it using ALT+. I checked and "UserInterruptHandler cmdDotEnabled" returns true in my image. Havent digged deeper. Can you please open an issue? Thx T.

Nov. 29, 2012