Scratch

SimpleScratch wrote:

" pasted it, selected it all, then right clicked on the text."
It was the selecting it that makes the difference - I hadn't done that.  JFI Its clicking on my scroll wheel that seems to bring up things on my RPi in squeak.
Ta again 

Yeah, on windows it is right clicking. On other operating systems it seems to be the mouse wheel.

SimpleScratch wrote:

"Every frame, the watcher resizes itself"  - Ah - I wonder if there are any other objects (timer.xpos,ypos etc) that do the same?

Simon

I think there is one other thing that might be doing something similar...

roijac wrote:

MathWizz wrote:

SimpleScratch wrote:

" pasted it, selected it all, then right clicked on the text."
It was the selecting it that makes the difference - I hadn't done that.  JFI Its clicking on my scroll wheel that seems to bring up things on my RPi in squeak.
Ta again 

Yeah, on windows it is right clicking. On other operating systems it seems to be the mouse wheel.

SimpleScratch wrote:

"Every frame, the watcher resizes itself"  - Ah - I wonder if there are any other objects (timer.xpos,ypos etc) that do the same?

Simon

I think there is one other thing that might be doing something similar...

lists are also pretty heavy...
if the stage resizes itself it'd be hilarious 

Lol. List ARE heavy. Also, when saving, they save the whole morph including all the red readouts. That is why lists save so slowly. I was just dealing with this issue here: http://scratch.mit.edu/projects/MathWizz/2741707 It took 10 minutes to save. xD

blob8108 wrote:

MathWizz wrote:

Lol. List ARE heavy. Also, when saving, they save the whole morph including all the red readouts. That is why lists save so slowly. I was just dealing with this issue here: http://scratch.mit.edu/projects/MathWizz/2741707 It took 10 minutes to save. xD

You should try Kurt! 

EDIT: not that I've implemented lists into the de/compiler yet... 

You mean I should edit the scripts with your exporter thing? You should make separate files for each list then. xD

Last edited by MathWizz (2012-08-22 16:05:23)

Oi - @blob8108 - don't distract MathWhizz from finding any other "watcher" thingies that need sorting out 
Simon

@MathWHizz
Members of the Scratch team have now got across the issue on the RaspberryPi forums - thanks once more for you efforts in getting this moving 

Simon
http://www.raspberrypi.org/phpBB3/viewt … 66#p158166

jji7skyline wrote:

The image probably won't work unless it's compiled to work with the ARM architecture as far as I know. Also, I'm not surprised that Scratch is slow, because RPi simply isn't a powerful machine 

The image has nothing to do with the ARM architecture. The Squeak VM has to be compiled for different architectures though.

Hello, Simon,
Updating variable watcher is high-cost operation, and Scratch do it every 100-150 ms.
So I made a patch to reduce update frequency to 1 sec.
http://dl.dropbox.com/u/1026229/reduceUpdateFrequency.1.cs
You can file in it from the "File List" tool after Shift-Click-R. To open "File List", click the background and choose "open..." from "World" menu then select "file list".

Cheers,
Kazuhiro Abe

P.S. I modified Scratch to handle GPIO of RasPi for controlling LEDs.
http://www.scratch.mit.edu/ext/youtube/?v=_Nf0Qh7tPCA
And I and my friends made a inexpensive sensor board named NanoBoardAG. It is compatible with PicoBoard and WeDo.
http://swikis.ddo.jp/WorldStethoscope/49
It works fine with RasPi.

SimpleScratch wrote:

Do you have a change file for your modification for controlling LEDs please? 

Here it is: RaspberryPiGPIO.cs
It enables to control a LED via GPIO of Raspberry Pi by motor blocks for WeDo.
It uses /sys/class/gpio/gpio4 pin for output.
You need to execute Scratch as a superuser.
No PWM support at the moment.

BTW, if continue this topic, it would be better to move to the following forum.
Connecting to the Physical World

Both the mods here make a huge difference to the performance of scratch on the raspberry pi. To the extent it is now useable without lots of sighing from young children!
I am also trying to use Scratch with the PiFace board. This communicated with Scratch using Mesh. As soon as i run the custom python scratch_handler for piface the CPU usage shoots up, although performance isn't too bad. Does the mesh in scratch have the same issue with checking for changes very regularly and hence hog the CPU?
I have put some delays in the python scratch_handler with no effect making me think the CPU usage is by scratch rather than the python script.

Anyone clever able to shed any light?

@dmurday - My scratch_gpio_handler.py code was originally taken from the pi-face handler back at end of July 2012.

The code then was quite a CPU hogger (one of the main culprits was jsut sitting waiting for ctrl-c to be pressed    )

But remember - this was just proof-of-concept at the time and the pi-face has only been released as a development board and not one intended to public (never mind young scratchers) to actually dailiy use.

I don't know if the code you have  has been updated but if you've a bit of time - you could maybe take my current code structure and try and graft the two together 
regards

Simon (simplesi/cymplecy)
PS AFAIK  Remote Sensor Connection is basically just Mesh and that doesn't actually seem to add in too much overhead 

Last edited by SimpleScratch (2012-09-26 13:31:06)

New to Scratch

This is the advanced section 

But seriously,  things have changed a lot in the last 6 months and Scratch has now been modified at source so nothing needed at the moment to adjust it

Simon