Re: [Phys-L] Physics "applets" in HTML5/JavaScript

[John Denker <jsd@av8n.com>]

On 03/10/2013 10:02 AM, Daniel V. Schroeder wrote:

> A decade ago, dozens--possibly hundreds--of us were writing Java 
> applets.  But support for this environment is disappearing very 
> quickly, and it never did work on mobile devices.

That's a good point ... and it has spawned an interesting
and informative discussion.  Thanks, everybody!

> Can anyone out there help me understand why there aren't more 
> educational physics "applets" written in HTML5/JavaScript?

> If you google for "html5 javascript physics" or something similar,
> you find almost nothing but pages about game development.

Well, it may not be quite thaaat bad, for a couple of reasons.

1) Not all of us have given up on writing physics applets.
 The google query mentioned above will not turn up all the
 relevant applets.  For example, I recently wrote
   http://www.av8n.com/physics/wavefunctions.htm#sec-animation
 which uses Javascript.  I ported it from an older Java thing.
 It uses somewhat-modern Javascript features.  As others have
 pointed out, the /canvas/ feature is crucial.

 Does this applet depend on HTML5?  I've never thought about
 it.  AFAICT it depends on semi-modern Javascript features
 but not any bleeding-edge HTML features.  In any case, 
 google has no clue about the HTML5 dependence (if any),
 so the aforementioned search will never find this applet.
 Maybe it will now, because of this email, but that's a
 fluke.

 I get 186 hits from the rather restrictive query:
  http://www.google.com/search?q=canvas+settimeout+fillstyle+physics+probability+energy
 and about 1440 from
  http://www.google.com/search?q=canvas+settimeout+fillstyle+physics+energy
 and 108 hits from
  http://www.google.com/search?q=canvas+settimeout+fillstyle+physics+energy+applet
 Those numbers are not huge, but they're not zero, either.

2) IMHO there is some nifty physics in my applet.  There
 are some lessons about the foundations of what we mean
 by probability (as represented by a scatter plot).  There
 are lessons about what Heisenberg will and will not let
 us do, when we start with an energy eigenstate and then
 ask questions about the position (which definitely does
 not commute with the energy).

 Also, I enjoy seeing quantitatively-correct righteous
 electron probability distributions, as opposed to the 
 kooky "artist's impressions" that you all-too-often 
 see in the literature.

3) I wouldn't be too disdainful of the game development
 systems.  The term "physics engine" is used in the game
 industry for a reason.  Guess how I know.  The point is,
 games need to model the physics, and the better games do
 a better job on the physics.

   http://en.wikipedia.org/wiki/Physics_engine

 Seriously, rather than writing a physics engine from
 scratch for research and/or educational purposes, I
 would strongly recommend you at least /consider/
 the idea of co-opting an open-source game platform. 
 I'm not saying it will be exactly what you need, 
 but there should be something like 80% commonality.
 That saves you a tremendous amount of work ... and 
 you get a better result in the end, including probably
 a better user interface than you ever dreamed of doing 
 by yourself.

 I am quite aware that it is more /fun/ to write your own 
 code from scratch, but if you care about productivity,
 efficiency, and the quality of the final product, you
 should seriously consider /not/ starting from scratch.

 Even if you wind up deciding that the other guy's physics
 engine is not good enough and you absolutely have to
 write your own from scratch, you will be better off
 because you will be smarter from having spent some
 time wallowing in the other guy's code.

 The best case is that you do such a good job that the
 game guys adopt your notion of what the physics engine
 should be like.  They help you finish it, extend it, 
 and support it.

 The worst case is that they aren't interested, and you
 wind up forking the code.  That's sad, but it's still
 better than starting from scratch.

 FWIW I know of a flight simulator program that has
 five different physics engines.  They're sufficiently
 modular that you can pick the one you want and plug
 it in.

==============

It's kinda sad what happened to Java.  It's not every
day you see something go from "favored" to "disfavored"
quite so quickly.

If they hadn't screwed up the security, it would have
been better than Javascript for any and all serious 
applications.  Strong typing leads to better code, 
and feature-rich /libraries/ lead to much much better
code.

When I ported the atomic scatter-plot applet, I spent
a couple of hours googling for a binary search utility,
discovering that the one I downloaded was no good, debugging
it, and eventually almost completely rewriting it.  In
Java or any other well-behaved language, I would have
just called a library routine and would have had every
confidence that it would be reliable and efficient and
well-documented.  The thing is only ten lines long, but 
that is ten lines of code that folks shouldn't have to 
write again and again.