[squeakland] [etoys-dev] Keyboard support

[Ricardo Moran]

2012/2/17 K. K. Subramaniam <kksubbu.ml at gmail.com>

> On Friday 17 Feb 2012 1:35:38 AM Ricardo Moran wrote:
> > It's now a single-key morph and it compares the keyStrings (I don't know
> if
> > this is good or bad). I hope it works across platforms (even if you have
> to
> > reconfigure the keys when you open the project). I only have windows here
> > so I would appreciate if someone can tell me if this works on Linux and
> > Mac.
> I tried it on Linux (Kubuntu 10.04). Single and double key presses work as
> expected. But third and further keys are ignored (i.e. up+left+right acts
> as
> up+left instead of up).
>

Yes, that happens to me also (in windows). I think that is a limitation of
the arrow keys, it works when using the WASD keys.


>
> KeyPressMorph gets my vote. It has awesome implications for interactive
> projects on handhelds and smartphone form factors. Thank you, Ricardo.
>
> A few observations:
>
> * apart from just isPressed state, could you also track timePressed?
> Authors
> can use longPress etc. for driving interaction. A key can also be used as
> an
> analog input (e.g. acceleration). Note that isPressed == (timePressed  >
> 0).

* If a user presses the button area accidentally, there is no way to escape
> out of it. How about clicking the button or the floating tip to cancel the
> operation?

* The tip "Press a key" is confusing. Suggest changing it to "Press new key"
> or similar and perhaps use blink or color change to indicate modal status.
>

Yes, I agree with these last three ideas.


> * The stringmorph "Key:" is superfluous. It comes in the way of grouping
> keys
> into a panel. This could be dropped.
>

I'm not sure about this one, though. I know it doesn't add any information
but I think the "Key:" serves as a title and a grabbing area for the morph.
If others agree with you, I'll drop it.


> * A new key is not initialized with a label. If a new key is initialized
> with
> 'a' label, 'currentKey != nil' becomes invariant. the check for nil in
> currentKey could be replaced with an assert or dropped.
>

You're right, actually there is no point in having a "keyButton" instance
variable.

Thanks, I'll fix these things.
Richo


>
> HTH .. Subbu
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.squeakland.org/pipermail/squeakland/attachments/20120219/6f7acf83/attachment.html>