[webkit-help] Background PNG decoding in WebKit

[Stephen Mattison]

Thanks for the reply Zoltan. What I am seeing on iOS is that sometimes the
decoding task runs in the main thread sometimes in the web thread. Is this
something that Apple is doing?

Cheers,
Stephen

On Wed, Oct 12, 2011 at 11:25 PM, Zoltan Horvath <zoltan at webkit.org> wrote:

> On Wed, 12 Oct 2011 22:53:25 +0200, Stephen Mattison <
> stephenmattison at gmail.com> wrote:
>
>  Hey Zoltan thanks for the input! Is there anything I can do to make sure
>> that the decoding always happens in the WebThread? Ie not adding to the
>> DOM
>> or creating a dom fragment or something?
>>
>
> Currently, there is no WebThread in WebKit. If you are using WebKit2 then
> there will be separated process for UI things and for Web things, what
> increases UI's responsiveness, but with the current implementation
> imagedecoding can block the loading a bit.
>
> Zoltan
>
>  On Wed, Oct 12, 2011 at 3:01 AM, Zoltan Horvath <zoltan at webkit.org>
>> wrote:
>>
>>  On Wed, 12 Oct 2011 03:39:19 +0200, Stephen Mattison <
>>> stephenmattison at gmail.com> wrote:
>>>
>>>  The problem I'm having is that it does not appear that PNGs are decoded
>>> in
>>>
>>>> the background in Webkit on iOS. Right now what I have is essentially a
>>>> ScrollView. I have a div the size of the viewport with another div that
>>>> scales with the content. I am scrolling the content div using a
>>>> transition
>>>> and translate3d to achieve GPU acceleration. The scrolling looks great.
>>>> However the content is composed of images and they cannot all fit in
>>>> memory
>>>> at once so I am loading them on demand. The problem arises when I set
>>>> the
>>>> background-image css property to a locally stored PNG image and then
>>>> attempt
>>>> to scroll. I get a noticeable delay/jerkiness/hiccup as the image is
>>>> decoded. Is there something I should be doing differently?
>>>>
>>>>
>>> Hey!
>>>
>>> WebKit is decoding images in the mainthread when there is new chunk of
>>> image and it hasn't decoded yet, thus it is imaginable that it blocks
>>> your
>>> main thread for 'longer' time.
>>>
>>> I have a sketchy implementation for Qt to support imagedecoding in a
>>> (currently 1) separate thread.
>>> If you are interested check: https://bugs.webkit.org/show_****<https://bugs.webkit.org/show_**>
>>> bug.cgi?id=69515 <https://bugs.webkit.org/show_**bug.cgi?id=69515<https://bugs.webkit.org/show_bug.cgi?id=69515>
>>> >
>>>
>>> Regards,
>>> Zoltan
>>> ______________________________****_________________
>>> webkit-help mailing list
>>> webkit-help at lists.webkit.org
>>> http://lists.webkit.org/****mailman/listinfo.cgi/webkit-****help<http://lists.webkit.org/**mailman/listinfo.cgi/webkit-**help>
>>> <http://lists.webkit.org/**mailman/listinfo.cgi/webkit-**help<http://lists.webkit.org/mailman/listinfo.cgi/webkit-help>
>>> >
>>>
>>>
>>
>>
>>
>
> --
> Using Opera's revolutionary email client: http://www.opera.com/mail/
>



-- 
Stephen Mattison
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-help/attachments/20111013/0a5839c5/attachment.html>