make getCanvasElementById safe, fixes #2

[hdgarrood]

Haven't tested this yet.

[paf31]

Looks good. If you can please test it, I can merge.

[hdgarrood]

Ok, I've tested it, and it worked for me in FF 33.0. I can do other browsers if you'd like.

[hdgarrood]

Also how do you feel about getCanvasElementByIdImpl being exported?

[garyb]

Looks solid to me, but yeah ideally getCanvasElementByIdImpl would be hidden.

[hdgarrood]

In that case, could you merge this and my other two PRs, and then I'll hide all the impl functions in one PR afterwards? Just so that I can do it all in one go.

[paf31]

Oops, just noticed, the type of getCanvasElementByIdImpl is a bit wrong. The way I usually type these things (not the only way by any means) is to say something like:

forall r eff.
  Fn3
    String
    (CanvasElement -> r)
    r
    (Eff (canvas :: Canvas | eff) r)

In the current type, the a is underconstrained.