Scratch

OldCodger wrote:

should BYOB not go the whole hog and fully adopt the functional programming paradigm?

In the course I teach for CS-major freshmen at Berkeley, we really try hard not to let our students become narrow adherents to any one programming paradigm; we visit functional, oop, and logic programming as well as a brief glimpse of client/server style with callbacks.  I love functional programming the best, but one of the great things about lambda is that it's universal, and you can build oop using it.

As a practical matter, we are aiming partly at high school teachers, most of whom have had most or all of their experience in the Java-based AP course and who therefore believe that oop is the only way to go.  We don't want to leave them out in the cold, although with time we hope to broaden their outlook.

So turns out Scratch 2.0 will be scratch with limited recursion, limited webcam, and limited cloud support. the scratch team at its best -.-
now that we have math functions, maybe add the (pretty essential) exponential one?
also noticed the NaN when the input is outside bounds (sqrt-1, acos91 etc.)    #whyidontlikejavascript
maybe you should kind of cache the categories, it takes so much time to switch them 
tried to profile it a bit, it looks like the biggest time-eaters are the String/TextMorph drawNew method 
if some optimization could be used...

joefarebrother wrote:

You know last time I checked, acos 0.5 = 60, NOT 60.000000000000000000001

Tell that to ECMA. I believe Snap uses Javascript's "Math" library, so any error like this is out of the control of Jens and bharvey.

As a future enhancement to Snap or BYOB could the windows be dockable/undockable so that users can rearrange them as desired?

Well, I think I'd like to be able to undock the stage and sprite windows from the editor because they are functionally different. The blocks window is too narrow at the moment to display some of the longer blocks in their entirety, and having free floating windows gives the greatest flexibility.

The Smalltalk development environment is an excellent model I think. Talking of Smalltalk as a model has any thought been given to alternative development methods such as a TDD option as the debug facilities are expanded.

will there be GUI components in 4.1, like in Java?

(....) The result being the same with Byob, my code was faulty. I have corrected it.
It gave me the opportunity to compare times for the same application :
Byob = 51 Snap! = 19 ....

I was wondering if you could make it so I can copy and paste into the text box on a block? I need it for the import to list block I've made.

Snap! alpha update:

Virtual keyboard and pinch zoom for iPads are now available in a first (still rough) version in the current alpha build.

Enjoy!

Last edited by Jens (2012-05-21 15:08:59)

@Jens
On my tab the drag-n-drop method is slow and touchy.
May I suggest you a different way to move palette block onto the script area ?
Select a palette block by clicking on it , then click on the position of the Script Area where you want to move it (no more dragging) .
Then you move the block inside the script area as usual.
Anyhow you have succeeded in making the proof of the concept .
Wonderful !!!!
I'm going now to make a test with a physical keyboard

Jens
I have used a Bluetooth to enter the numerical values and I've written my first Snap! tab appli.Even WARP works.
I've not yet found a Screen capture application, so I've taken a picture of the screen and sent it to your email.

Hi Graham and Xavier,

thanks for testing the new features for tablets!

It's a bit tricky to distinguish my way to pop up a context menu (tap one finger on the object, hold it and then tap another one anywhere) against the device's multi-touch pinch gestures. That's why you will frequently get a context menu if you pinch anything, and things will appear not react to pinches occasionally. I'm still trying various ways to avoid this and I'll let you know when I change anything.

Invoking the virtual keyboard for a purely canvas based application is unbelievably kludgy, but at least it seems to work now. Again thanks for testing!

At the moment, serialization is incomplete and partly broken. The rotation style options do work, but they don't get saved, the same applies to any scaling. Worse, there are serious flaws regarding global variables and local stage variables. Again, they work fine, but once you save/reload a project, there are errors. We're working on that one...