[Sugar-devel] webkit, hulahop; developing apps using browser engine DOM for widgets

[Lucian Branescu]

I don't think I have the results anymore, but benches between
epiphany-webkit and epiphany-gecko were very similar.

The benchmarks I've used stress the browser engine, especially
javascript. Perceived performance is usually better with webkit as
well, though.

2009/6/10 Jonas Smedegaard <dr at jones.dk>:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: RIPEMD160
>
> I wrote earlier:
>>> Politics aside, I do not doubt that webkit might perform better than
>>> mozilla.  In some situations.  Optimized in certain ways.  'Cause
>>> there are a bunch of complex factors, as I understand it (and I don't
>>> understand it that deeply, really).
>
> On Wed, Jun 10, 2009 at 03:22:37PM +0100, Lucian Branescu wrote:
>>Before GSoC started, I did my own tests of webkit vs gecko. Firefox
>>lost everywhere and lost especially bad on memory usage.
>>
>>http://dl.getdropbox.com/u/317039/webkit%20vs%20gecko%20osx.txt
>>http://dl.getdropbox.com/u/317039/webkit%20vs%20gecko%20soas.txt
>>http://dl.getdropbox.com/u/317039/webkit%20vs%20gecko%20winxp.txt
>
> Did you simply test default configurations of each tools, or is the
> optimizations documented somewhere?
>
> As I understand it, especially the large tools ike Mozilla can be
> optimized for different purposes.
>
> Example: prefetching pages (default enabled in Mozilla) is probably bad
> on memory and has no benefit in load-a-bunch-as-fast-as-possible tests
> but improves perceived load time in load-pause-load scenarios (i.e.
> normal web browsing).
>
> I can also imagine how image buffering hurts memory but affects e.g.
> scrolling performance.
>
>
> So it really makes little sense to me to compare default settings of
> these tools, if the purpose is to find the best performance in netbooks:
> Mozilla defualt are most likely optimized for more powerful hardware.
>
> Either investigate options to tune and document all optimizations, or
> compare browsers (based on each of the core frameworks) that are
> optimized for low-ressource environments.
>
> The gecko-based Kazehakase is (according to its author, which I had the
> opportunity to meet in person when invited to Japan a few years ago)
> leaner than Firefox, but still targeted normal desktop environment.
> Newer releases of it seems to support switching between gecko and webkit
> as backend.  Perhaps that is a good candidate for head-to-head
> comparison of similarly targeted tools.
>
> Epiphany seems to also support both gecko and webkit backends.
>
> Sorry, I have no suggestions for ideal tools to compare for this
> challenge - tools that are optimized for reduced memory footprint while
> still being targeted normal X11 environment.
>
>
>  - Jonas
>
> - --
> * Jonas Smedegaard - idealist og Internet-arkitekt
> * Tlf.: +45 40843136  Website: http://dr.jones.dk/
>
>  [x] quote me freely  [ ] ask before reusing  [ ] keep private
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.9 (GNU/Linux)
>
> iEYEAREDAAYFAkov8bQACgkQn7DbMsAkQLhYRwCfXNu2CWy2Cw7HN0izXJncyphk
> 8H8AmwX/FCkBIgCdzkSrtml6xLrc9Hw2
> =QFId
> -----END PGP SIGNATURE-----
> _______________________________________________
> Sugar-devel mailing list
> Sugar-devel at lists.sugarlabs.org
> http://lists.sugarlabs.org/listinfo/sugar-devel
>