bharvey wrote:
fullmoon wrote:
roijac wrote:
...
Out of curiosity, where did you find a block called THE BLOCK WITH INPUTS?
He Built His Own Block, of course!
Oh! I forgot you could do that!
I usually think in terms of anonymous closures rather than formal function definitions, so that possibility actually slipped my mind. Apologies!
Offline
fullmoon wrote:
bharvey wrote:
fullmoon wrote:
Out of curiosity, where did you find a block called THE BLOCK WITH INPUTS?He Built His Own Block, of course!
Oh! I forgot you could do that!
I usually think in terms of anonymous closures rather than formal function definitions, so that possibility actually slipped my mind. Apologies!
Offline
Hey, bharvey. How are you recovering?
Offline
ProgrammingFreak wrote:
Hey, bharvey. How are you recovering?
Thanks for asking. The original intestinal problem is all better, and if all goes according to plan, tomorrow the surgeon will declare my surgery wound healed so I can vaccuum my floor again.
Offline
xly wrote:
@Jens :
Your SNAP works perfectly well, see my :
http://www.xleroy.net/ByobTuto/New/squarespiral.gif
To fill values in BLUE instructions is still tricky (I am using Firefox)
Why couldn't you do
[blocks]
<go to x -50 )y 150
[/blocks]
?
Also, you know how to make the hyperlinks in BBCode? It saves a couple seconds for readers. xP
Why is "if" in green like that?
And just to say, FF5 is better.
Jens, when do you think stop all can have the stop sign on it?
The variable is a little too long. Yeah, I'm referring to the empty space at the end.
In the block, is it supposed to update the color when the mouse is up?
Offline
Thanks, Xavier, and scimonster for your testing and feedback! I love the squiral example and keep a similar running in a forever loop on my desktop all day.
Which version of Firefox are you currently using? (v.6 or even beta 7 should be pretty fast and also less memory intensive). Yeah, I need to decrease the minimum width for reporter blocks, so short ones don't have a lot of empty space. Icons for the stop block and for the turn blocks are also on my list, but they're easy to do, so I'll concentrate on more hard stuff for now (script variables are next).
Thanks again!
Offline
Ooh, script variables!
There's V6? I knew I should update.
Offline
bharvey wrote:
ProgrammingFreak wrote:
Hey, bharvey. How are you recovering?
Thanks for asking. The original intestinal problem is all better, and if all goes according to plan, tomorrow the surgeon will declare my surgery wound healed so I can vaccuum my floor again.
Thats great! I'm glad to here that you're getting better.
Offline
Hi Xavier,
I'm very glad that you're now experiencing better performance with the new prototype, as Brian keeps pointing out, that's the main reason for the reimplementation. The HTML5 canvas API is still very new and there have been spectacular performance gains in almost all recent browser versions. This makes me confident that BYOB will be alright in JavaScript, even though Squeak technically is (much) faster.
Offline
Jens wrote:
The HTML5 canvas API is still very new and there have been spectacular performance gains in almost all recent browser versions. This makes me confident that BYOB will be alright in JavaScript, even though Squeak technically is (much) faster.
... which makes me nervous, because the performance I care about has nothing to do with graphics rendering, but with CPU-heavy function crunching.
Offline
@bharvey&Jens
See this example : http://www.xleroy.net/ByobTuto/New/geometricspiral.gif
This applications takes 4mn 50 with Byob and 1 mn 09 with Snap. It is a significant progress. But anyway nothing to compare with tremendously fast XLogo interpreter written in js. See 3D examples at :
http://xlogo.tuxfamily.org/fr/html/examples-fr/3d.html
(each example can be run - see for example "eponge de Menger" )
Offline
Hi Xavier, Snap will get much faster in its atomic blocks. For a preview you can open an inspector on the Sprite in the web prototype (right-click on the pointer thing, select "a Sprite Morph" and then "inspect", and type the following JS code into the evaluation pane (at the bottom):
this.startWarp();
this.down();
this.tree(12, 120, 20);
this.endWarp();
then select all (ctrl-a), right-click on the selection and choose "do it". This will draw a LOGO tree with 12 iterations, and pretty fast, too. It is about the speed which atomic blocks will have in Snap, if we decide on "brutal atomicity" (which is yet a matter of experiment and debate)...
Offline
@Jens - "Nascituri te salutant"
I've tested your "tree" as mentioned above.
Wouahh !!! That makes a change.
After that I have used the following function to make time comparisons:
SET CONT 1
SET TURN 30
REPEAT TURN *360
[MOVE 1 STEP
TURN 1500/CONT
CHANGE CONT BY 1]
The results !!!
Byob Without Turbo Mode = 290 s
Snap = 100 s
Byob +Turbo Mode = 8 s
No comment !
I guess that your "tree" also uses something similar to turbo mode (atomic).
Offline