Re: Transport change in MS Proposal

On 9/3/2012 12:36 PM, Cullen Jennings (fluffy) wrote:
> On Aug 29, 2012, at 3:35 PM, Martin Thomson <martin.thomson@gmail.com> wrote:
>
>> Javascript should not block in this way.  If your javascript does not
>> return control to the browser at least 60 times a second,
> Have you actually looked at how long the JS blocks on most pages that use a bunch of it? I think you are in the wishful thinking zone on this one.

Mozilla has a major project that's been running around a year called 
'Snappy' (https://wiki.mozilla.org/Performance/Snappy)

We have a *goal* of <=50ms responsiveness when typing in a textbox, and 
60fps animations in UI operations.  We've done major surgery on our 
GC/CC code to greatly reduce pauses due to them, but you can still get 
pauses over 50ms if you're a tab hoarder, among other reasons.  And then 
there's website/app JS code, like gmail, games, etc - how long can they 
control things, causing your WebRTC application to wait?

If you can get off the Main Thread (onto a worker, if you expose these 
APIs to workers), things are a lot better - but may need more locking 
and synchronization logic.

-- 
Randell Jesup
randell-ietf@jesup.org

Received on Tuesday, 4 September 2012 05:05:32 UTC