[Sugar-devel] testing composition for speeding up graphics performance

[Gary C Martin]

Hi Eduardo,

On 21 Jun 2009, at 23:46, Eduardo H. Silva wrote:

> 2009/6/19 Gary C Martin <gary at garycmartin.com>:
>> On 18 Jun 2009, at 10:29, Tomeu Vizoso wrote:
>>
>>> On Thu, Jun 18, 2009 at 11:21, Sascha
>>> Silbe<sascha-ml-ui-sugar-devel at silbe.org> wrote:
>>>> On Thu, Jun 18, 2009 at 11:09:06AM +0200, Tomeu Vizoso wrote:
>>>>
>>>>> Wonder if that's intended, as going to a zoom level involves some
>>>>> kind
>>>>> of animation, i.e. is not intended to take you there as fast as
>>>>> possible.
>>>>
>>>> What kind of animation? I can't remember ever having seen one
>>>> (unless you
>>>> count the time it takes to populate the screen).
>>>
>>> It's not supposed to be too obvious. It's more visible if you go  
>>> from
>>> the home level to the neighborhood, you should see the XO zooming  
>>> out
>>> before the screen is redrawn.
>>
>> Yes I agree with that, but only if you are changing from one of the
>> base 3 zoom levels to another. Once you switch to an Activity level
>> zoom, there should be no zoom from some old negihbourhood/group/home
>> state to a new one that you're trying to switch to. If we had posh HW
>> accelerated compositing, the animation should be something like a
>> cross-dissolve and dolly back from the Activity view level to the new
>> negihbourhood/group/home view.
>>
>> Of course, the current behaviour is correct 33% of the time :-)
>> because you may be going back to the same zoom view as before, it's
>> only the other 66% of the time that you get shown an initially
>> incorrect zoom view.
>>
>> Hmmm, I seem to remember an old ticket where this problem was showing
>> up before, had some almost complete patches if I remember (gee, now
>> who was working on that, I remember testing the patches through
>> several revisions)... Any one know?
>
> Is it this one?  It doesn't have patches though
>
> http://dev.laptop.org/ticket/4039

Oh, thanks for reminding me. No, this is not the ticket I remember  
(though very closely related to the same issue), I'm sure there were  
several cycles of patches attached to the trac ticket I tested. I'll  
try and have  a look for it as well.

Regards,
--Gary