InputEventSensor and polling

Hi all, 

I've been working on Phratch, a port of MIT Scratch to Pharo, mostly to learn Smalltalk and also because Scratch is a great learning environment and I can't bear to see it end up in Flash.

So I've been digging about in the old code to fix bugs and I'm getting a sense that there's some conflict between the way the Scratch team got things working in Squeak, and the way the Pharo team things want things done. 

I must say I think the Pharo team have things right in their drive for a cleaner, simpler system, however that means I'm a bit stuck.

See the attached image of my Scratch script.

Here's where I am...

1. Keyboard events for 'When <key> pressed' HatBlockMorph (with then pen commands) work fine as these follow the regular Morphic event model

2. But 'if <key> pressed' operation (with the stamp command inside) doesn't work because it relies on polling the InputEventSensor, and I think this bit of this class doesn't work as I expect.

My view right now is that for 2, I need to use some kind of polling, and InputEventSensor has methods such as keyboardPeek and peekKeyboardEvent that imply you can poll,

but looking at the implementation, which uses WaitfreeQueue I can see that calling these methods results in events being consumed, hence my post to Stack Overflow...

http://stackoverflow.com/questions/16395418/cant-peek-keyboard-events-multiple-times-in-smalltalk-pharo-2

I've hacked around to come up with something that sort of works, outlined in this post:

http://stackoverflow.com/questions/16495436/saving-modifications-to-existing-kernel-methods-to-monticello

...but I get the feeling I'm doing the wrong thing here, and that I should come up with a solution that better fits the 'Pharo way'.

So, let me know what you think. I'm happy to provide more info and to research better solutions, let me know...

Finally, apologies for the long post! 

Best regards, 

-Eric.

--
Eric Clack
[hidden email]
East Sussex, England.

<keypress.gif>

But as far as I can see, that "if key <space> pressed?" cannot be event driven, because we don't want to start execution at that block if the space key is pressed, rather we want that to evaluate to true if the space key has been pressed recently ("recently" seems a problematic term to use here but that's another issue!).

Here's the code I started throwing together to make this work -- see the bit following this line:

    type = EventTypeKeyboard

InputEventSensor>>processEvent: evt 

"Process a single event. This method is run at high priority.

The event record is:

<type><timestamp><character code><updown><modifier keys>...

where updown is:

 0 - keystroke

 1 - key down

 2 - key up

NOTE: You must ensure that there is an instance variable keyPressed."

| type updown |

type := evt at: 1.

"Treat menu events first"

type = EventTypeMenu

ifTrue: [

self processMenuEvent: evt.

^nil].

"Tackle mouse events first"

type = EventTypeMouse

ifTrue: [

"Transmogrify the button state according to the platform's button map definition"

evt at: 5 put: (ButtonDecodeTable at: (evt at: 5) + 1).

"Map the mouse buttons depending on modifiers"

evt at: 5 put: (self mapButtons: (evt at: 5) modifiers: (evt at: 6)).

"Update state for polling calls"

mousePosition := (evt at: 3) @ (evt at: 4).

modifiers := evt at: 6.

mouseButtons := evt at: 5.

^evt].

"Finally keyboard"

type = EventTypeKeyboard

ifTrue: [

"Update state for polling calls"

modifiers := evt at: 5.

updown := evt at: 4.

(updown = 2) ifTrue: [ 

keyPressed := nil

] ifFalse: [  

keyPressed := evt at: 3 ].

^evt].

"Handle all events other than Keyborad or Mouse."

^evt.

Then I use this:

InputEventSensor>>keyPressed: asciiValue

"Is this key being pressed?"

self nextEvent.

^keyPressed = asciiValue

So should I be doing this? Should I test out these changes and suggest a patch to Pharo? I'm happy that the answer might be "no", but what can I try instead?

Thanks for your help.

-Eric.

--
Eric Clack
[hidden email]
East Sussex, England.