Re: (X)HTML5 + SMIL?

David,

Thanks for pointing out also my Timesheets implementation. I haven't been working on it for a while. However, if I remember correctly I was able to implement all the major features in the SMIL Timesheets document [1]. Even, prefetcing and animations work [2]. Implementing the missing features [3] shouldn't be a big problem

The layout is taken care by the browser CSS layout engine, which is pretty efficient in most modern browsers. The Javascript part is rather simple, and thus I didn't find any real performance problems.

I hoping to see some real life use cases. Then, it would be possible to do more detailed performance analysis using, e.g., Firebug [4].

Yours,

Petri Vuorimaa
Aalto University

[1] http://www.w3.org/TR/timesheets/
[2] http://www.tml.tkk.fi/~pv/timesheets/tests.xhtml
[3] http://www.tml.tkk.fi/~pv/timesheets/features.xhtml
[4] http://getfirebug.com/

On 8.3.2010, at 12:06, David Leunen wrote:

>  
> I'm curious what your experience implementing this in JS has been, whether there are particular performance issues you've encountered, or things that were particularly difficult that you may have had to hack around or anything?
> 
> 
> I tried to introduce the least hacks possible. Nevertheless, I consider the entire script is a hack.
> I haven't stress tested it, so I don't know where are the performance bottlenecks. I guess it depends on the JS engines.
> The main difficulty is implementing additivity. The script need a major rewrite for that (i.e it doesn't strictly follow the model SMIL defines : aka sandwich model).
> Another bug is that animations' starts are not strictly calculated, but relies on events. So they may become out of sync, if they are repeated many times.
> Apart from that, everything was pretty straightforward, if I remember well.
> 
> 
> (There's also that timesheet implementation : http://www.tml.tkk.fi/~pv/timesheets/ )