[Sugar-devel] ideal flash activity to be remade as karma activity

[Lucian Branescu]

I did some more testing, this time in SoaS inside VirtualBox. It's as
close as I can get to an XO.

I used this code for hulahop
http://wiki.laptop.org/go/Sugar_Code_Snippets#WebView
Similar code for pywebkitgtk.

Results here http://dl.getdropbox.com/u/317039/webkit%20vs%20gecko%20soas.txt
They seem to be consistent with my results on OS X.

Also, it occured to me that making a pywebkitgtk backend for hulahop
would be easier than I had thought.

2009/3/28 Lucian Branescu <lucian.branescu at gmail.com>:
> I'm such an idiot, I forgot to link the results.
> Here http://dl.getdropbox.com/u/317039/webkit%20vs%20gecko.txt
>
> 2009/3/28 Lucian Branescu <lucian.branescu at gmail.com>:
>> I've been doing some benchmarks between firefox and safari nightlies
>> on OS X Leopard x86. Firefox had Tracemonkey and Safari had
>> SquirrelFish Extreme.
>> Every test was run three times and the best time was recorded. Memory
>> usage is Real Size as reported by the OS X (Activity Monitor and top).
>>
>> I asked around on #firefox and #webkit, people in general agreed that
>> the v8 bench is purely synthetic, since it mainly tests recursion and
>> memory management. SunSpider was said to be the most indicative of
>> real-world usage and Dromaeo the most comprehensive.
>>
>> The results may vary greatly on linux, and in particular the XO.
>> bernie (or bemasc, sorry I keep getting your nicks mixed up) was
>> thinking of testing on the XO, I don't know how far he got. (if you do
>> test, also try the chrome experiments)
>>
>> In the end, I believe it may be beneficial for Sugar to switch to
>> webkit, but it would be best to not introduce a new dependency only
>> for the web-activity thing. The difference isn't that great and things
>> like Bespin manage to work fine on firefox.
>>
>> A webkit backend for hulahop would probably be an interesting project,
>> but I may be wrong since I don't know that much about hulahop's API.
>>
>>
>> 2009/3/27 Jameson Quinn <jameson.quinn at gmail.com>:
>>>
>>>>
>>>> 3 months is a very short period to bring not only 2 but 4 people to work
>>>> together across multiple time zones.
>>>
>>> I brought this up on #gsoc yesterday, and LH (the top authority) was
>>> somewhat skeptical about any coordination. I said we would leave it up to
>>> the students involved, and she said that sounded good, but Google would have
>>> to approve any final plan, and they wanted to avoid any dependencies,
>>> whether implicit or explicit, on the future work of anyone besides the
>>> mentor and the student.
>>>
>>> Jameson
>>>
>>>
>>
>