On Tue, Sep 11, 2012 at 10:34 AM, Simon Pieters <sim...@opera.com> wrote: > On Tue, 11 Sep 2012 00:14:25 +0200, Glenn Maynard <gl...@zewt.org> wrote: >> If you really want to protect users from the behavior of pages, you'd >> really need to make creating the context cheap. > > This was exactly my reaction as well. Introducing a new method doesn't solve > the problem. There will still be pages that create the context and then > don't use it. Browsers just have to suck it up and be fast anyway, just like > with all other features on the Web. > > I think we shouldn't add supportsContext() since it doesn't provide authors > with new information and browsers will have to do the above anyway.
I actually like Ashley Gullen's proposal (canPlayType-inspired) because it brings consistency to the platform and offers a general pattern on how to test for the presence of features which can't be asserted for certain without costly lookups. navigator.onLine property is another example where a similar pattern would make (more) sense (than the current boolean returned instead). Basically, this is trading performance for the risk of false positives. --tobie