Scratch

Hardmath123 wrote:

EDIT: Maybe I'll write a JS library to do stuff like that:

Code:

ITER([1,2,3,4,5,6,7,8,9,10], function(num) {
  alert(num*3);
});

Shouldn't be too hard.

It's called [].forEach.

Jens wrote:

Okay, here is the thing: When setting a variable, if the value can be interpreted as a number (e.g. if it is a string beginning with a number), Snap will cast it to one. Numbers take precedence over strings. In your example you can prevent this from happening by adding an alphabetical character at the beginning of the word.

That's not the correct behavior. Operations that expect a number should cast their operands to numbers right before they use them (using the expression `+x || 0`—not parseInt(x), which does weird things with stuff like '0x123' or '123abc'). Variables can hold any string; the SET block takes a string as its second argument and should not convert that argument to a number.

joefarebrother wrote:

Jens wrote:

Okay, here is the thing: When setting a variable, if the value can be interpreted as a number (e.g. if it is a string beginning with a number), Snap will cast it to one. Numbers take precedence over strings. In your example you can prevent this from happening by adding an alphabetical character at the beginning of the word.

That's what happens in scratch, too. I've found a very useful purpose for it in my scheme interpreter i'm writing in scratch.

Scratch casts strings to numbers with asNumberNoError (which is exactly equivalent to `+x || 0` in JavaScript) only when it requires a number, e.g., in the implementation of <() + ()>. It never casts strings to numbers when they're used as inputs to a block which takes a string, e.g., in the implementation of <set () to ()>. Try this simple script:

<script variables (var)>
<set (var) to (123abc)>
<say <var>>

In Scratch, the sprite says "123abc". In Snap!, the sprite says "123". Scratch is right. Snap! is wrong.

Last edited by nXIII (2013-01-20 16:09:09)

I changed the SET type coercion. My test cases all work fine with it, but there is a really cool CS10 student project which doesn't work right anymore. Therefore I'd appreciate some more testing, if anyone has got time at their hands. Do your experimental Snap! projects still work as they used to / as you'd expect them to?

Thanks!

Which browser and OS do you notice this anti-aliasing behavior in? Is this something new (the rendering hasn't been changed for ages)

There should be a way to specify that you should do something when a user-created data structure is garbage collected.

joefarebrother wrote:

There should be a way to specify that you should do something when a user-created data structure is garbage collected.

Implement your own garbage collection. 

Hmmm. I'm using the same configuration and it doesn't produce any problems. When did you last reboot? 

(Heheh, this used to be the antidote to whatever Windows was doing wrong. Turns out rebooting often is a remedy for strange Mac behavior, too).

I don't know. Anyway, it's gone now. 

Third time I'm bumping my JSON idea. This time, I've got a reason you've got to be interested in.    I'm writing a raytracer in Python, and yesterday night I had a flash of inspiration and implemented triangle primitives (I'm trying to derive all the math I use so that I actually learn something). I wanted a cool little demo, So I thought I'd make a 3D "S" in Scratch-style font. I basically aimed to do what I did here; except raytraced. 

I wanted to write myself a simple little GUI to do it (inputting a large number of 3D coordinates manually is painful!), so I tried to use Snap!. The problem was that I didn't have a good way to get the coordinate data from Snap! lists.

If we had JSON support, I would simply dump the list into a JSON string and open that as a datauri in a new tab (ok, you also need an "open [] in new window" block). But since we don't I tried to write a ist-to-JSON block within Snap!. And crashed Safari. 

Convincing? How about just an "export []" block which turns the input whatever type it is, into JSON and opens it in a new tab? 

@Jens Whoa, 3D printing? How'd you do that?

What's CORS?

joefarebrother wrote:

What's CORS?

Cross-origin resource sharing

Hardmath123 wrote:

Convincing? How about just an "export []" block which turns the input whatever type it is, into JSON and opens it in a new tab? 

You can write your own pretty trivially, and just copy the result into a file somewhere.

EDIT: Oh wait…Morphic doesn't support copy-paste. You can use that bookmarklet thingy I made a while ago.

EDIT2: @bharvey your ring has quite a following 

Last edited by nXIII (2013-01-22 20:31:38)

Hardmath123 wrote:

nXIII wrote:

EDIT: Oh wait…Morphic doesn't support copy-paste.

Exactly. There's no way to export or import a large amount of data. I guess I'll have to make that server after all... 

javascript:var d=document,s=d.createElement('script');s.src='http://dl.dropbox.com/u/10715865/Web/Snap-Old/inject/native-editors.js';d.querySelector('head').appendChild(s)

It's a little buggy/outdated, but it should do fine if you need to copy-paste. (Run it on startup so that all new textbox morphs are native, or even better include it in a local copy to catch the ones created at startup)

Last edited by nXIII (2013-01-22 21:14:40)

Jens wrote:

xly wrote:

@Jens
In the good old Byob3 days we had a feature, where by right-clicking on a list display window we had the two commands : Export, Import allowing to create and read text files. I've used this to export/import lists of coordinates !

Yes, this is clearly The Right Thing. Only that in Snap we're using generic variable watchers for lists, so we want to import plain text files directly into a variable's value, and do the parsing ourselves.

Check out today's version: I've added "import..." and "export..." entries in variable watchers' context menus. This lets you get text data in and out of Snap! in a hopefully simple way.

With Firefox I get only the Import feature
With Chrome, nothing, neither Import nor Export.
I'm not qualified to give you any advice, but, if the Imported file is a text file, each of its line could be an item of a named list (as for Byob)  Then up to the user to parse each item of this list accorgingly. I've done this for coordinates where + and - were converted into  and m and numeric coordinates into strings.