flex-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rick Winscot <rick.wins...@gmail.com>
Subject Re: Apache Flex suggestion - dumping SWF support in favor of HTML5 - listen to Steve
Date Mon, 06 Feb 2012 18:42:17 GMT
Doug - I agree with your bottom line... trajectory is _very_ important. However, I think that
the importance of any trajectory is directly dependent upon the quality of the samples. In
this case, I'd have to say that this trajectory is grossly misleading... I'll give you several
reasons why.

First, the Flash security model is stable, mature, and consistent across all browsers. The
figures for Flash include the security overhead for managing crossdomain, mouse / touch /
keyboard interaction, and whatnot. Adding this kind of capability into JavaScript... will
likely impact execution time ( It always does ).

Second, is the lack of true 1:1 comparisons between technologies. For example, if we want
to validate the performance of runtime asset management - we could use the Loader in ActionScript
and in JavaScript... hide / show a div tag? 

Third, the gaps in browser performance are significant ( in the statistical sense ). In order
to consider this trajectory valid... the standard deviation between browsers would have to
be small; very small ( http://www.webkit.org/perf/sunspider/sunspider.html ).

Fourth, a plateau is a 'signature' of stability; that a point of diminishing return has been
identified. When you get to this point it's wise to conserve efforts for innovation / paradigm
shift ( what takes you to a new plateau ). Spending countless hours trying to squeeze a little
more performance out of operation X... is difficult to justify. The move to GPU for Flash
is a good example of just such an innovation / paradigm shift. So... IMO, a plateau isn't
a sign of weakness - it's a sign of maturity. I would feel more comfortable with JavaScript
as a solution if it had a few plateaus under it's belt. ;-)

Rick Winscot

On Sunday, February 5, 2012 at 4:03 PM, Doug McCune wrote:

> > 
> > http://blogs.adobe.com/**avikchaudhuri/2012/01/17/the-**
> > v8-myth-why-javascript-is-not-**a-worthy-competitor/<http://blogs.adobe.com/avikchaudhuri/2012/01/17/the-v8-myth-why-javascript-is-not-a-worthy-competitor/>
> > 
> I'd argue the important thing is not the current delta between JS and AS
> performance. For rendering-related tasks ActionScript is still way ahead
> (as that blog post tries to highlight and praise). However, that's not what
> you should focus on. Take a look at this chart:
> http://iq12.com/blog/as3-benchmark/ which shows the incremental speed
> improvements for AS code execution vs JS code execution.
> Ignore the numbers in the chart, and ignore the comparison of AS to JS. The
> benchmark only highlights things where JS excels (non rendering things).
> Focus on only one thing in that chart: the number of times the lines
> change. Since 2007 (when FP9/AS3 came out) there are 3 times when AS3
> performance increased. Compare that to the line for JS performance in
> Chrome, which has 8 jumps in performance since 2009. That's the difference
> that matters.
> And yes, you can argue that the GPU stuff with Molehill should be taken
> into account, etc (although that doesn't help with performance of any
> content not specifically written for GPU rendering). But I think that chart
> tells me more than anything else in this debate. AS3 performance has
> stagnated. JS performance has consistently increased. Is it as good as AS3
> right now? No. But largely that doesn't matter. It's the trajectory that
> matters.
> But enough about JS performance. My point wasn't to talk shit about Flash.
> My only point was that writing off HTML/JS as inferior is naive and
> dangerous.

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message