Scratch

This topic was made in the same way as yours; someone took the initiative and created one (which is awesome by the way). But if Brian or Jens create it, they'll be able to manage it and keep it up-to-date. Unless I'm mistaken, they also asked us to wait until the official Snap! release. That way the new topic isn't plagued with many pages of pre-alpha and beta testing.

Last edited by shadow_7283 (2011-11-24 20:31:16)

joefarebrother wrote:

Read the rest of the code. It uses the (copy of()) block, which creates a new list as a copy of an existing list.

Ah, okay, sorry.  That's exactly how the BYOB 3 tool blocks do cons and cdr.  But COPY OF is only a constant factor in speed; it still takes time proportional to the size of the list to do the copying.  The new primitive CONS and CDR blocks are constant time, not linear time, so they make a huge difference for large lists.

fixed!

Thanks, Xavier, could you try your factorial example again and let me know if it still doesn't work for you?

Hardmath123 wrote:

No and/or/not? I need it for a Mandelbrot set plotting.

The rest left as an exercise for the reader. 

Jens wrote:

fixed!

Thanks, Xavier, could you try your factorial example again and let me know if it still doesn't work for you?

OK, It's fixed. Good relief !!
In the meantime I've tested the "fern" application the difference
With/Without J. Malone modification:
490 sec/870 sec
It's quite impressive !
Warp time is roughly the same, 6,80 sec

Or we could just change the name of this topic.

henley wrote:

Or we could just change the name of this topic.

No, the idea is that a year from now, someone can find the Snap! topic and the very first thing in it will be the release announcement, and then all the program examples after that will be in the final UI form, etc.

bharvey wrote:

Less than a minute for N=100.  Predicted less than 10 minutes for N=1000.

Not bad.... will we see a further dramatic speed increase? If it could just be pushed a bit further, I think Snap! would attract a lot of people, maybe even some of the Scratch 2.0 crowd (not that it's a competition or anything).

shadow_7283 wrote:

will we see a further dramatic speed increase?

At some point we plan to compile the scripts into JS, at least the ones that define blocks.  Probably not before the beta release, though.  Also I want to do tail call elimination at some point.  Until we do those things, my guess is that any speedup will be incremental rather than huge.

EDIT: except for a few UI known slowdowns, such as clicking OK after changing the name of a variable.  That'll get looked at soon.

Last edited by bharvey (2011-11-27 14:44:46)

Byob is the best 

djdolphin wrote:

Are you going to add the green flag block in?

Yes, of course, all the missing features will be in.  We've been concentrating on the bare minimum of features to concentrate on the core of the UI and the core of the evaluator.  But we're now both convinced that we can run large programs quickly, so you'll start to see missing pieces appear soon.

jurk wrote:

there should be a BYOB website like scratch so users can share code and projects

That takes a long time to make. It would be extremely hard for two people to design an entire copy of the Scratch website, but you could compile projects, and post them to the forums.

henley wrote:

jurk wrote:

there should be a BYOB website like scratch so users can share code and projects

That takes a long time to make. It would be extremely hard for two people to design an entire copy of the Scratch website, but you could compile projects, and post them to the forums.

Ok I was just wondering I know it takes a long time to make but wouldn't be cool if you made one in your spare time? like just a quick won with just forms and a home page.