Hey Simon,
I like your proposal! :)
I think we should add ENABLE guards around qtimagedecoders (enabled by
default), and modify the code to support webcoreimagedecoders at the same
time. Thus people can share their observations when they tried it out,
then if everybody agrees on the switching we can easily remove
qtimagedecoders completely.
I'll gladly make patches for the switching.
Cheers,
Zoltan
On Fri, 10 Feb 2012 11:21:41 +0100, Simon Hausmann
<[email protected]> wrote:
Hi,
I'd like to propose that we switch over to using the WebCore image
decoders.
Zoltan has spent some time comparing performance and memory usage
between the
ImageDecoderQt (and the qt plugins for image decoding) and the WebCore
image
decoders, in https://bugs.webkit.org/show_bug.cgi?id=71555
My interpretation of the result is that performance and memory usage wise
there's not much of a difference. However the WebCore image decoders
have some
advantages that are of a different nature:
1) They are an actively maintained piece of code. Unfortunately the
Qt
image decoder plugins are subject to less tender loving care. The WebCore
decoders are used by pretty much all other ports, and from my totally
biased
point of view we have also found them working great for the N9 with the
Qt
based browser/webkit.
2) They support image downsampling, which is particularly nice for
mobile
devices with less memory than our gigabyte filled workstations.
3) The jpeg decoder supports progressive/interlaced jpegs, which is
again
an nice thing for image loading over slow connections.
Therefore I'd like to re-open
https://bugs.webkit.org/show_bug.cgi?id=32410 .
Any opinions / objections / thoughts?
Simon
_______________________________________________
webkit-qt mailing list
[email protected]
http://lists.webkit.org/mailman/listinfo.cgi/webkit-qt
_______________________________________________
webkit-qt mailing list
[email protected]
http://lists.webkit.org/mailman/listinfo.cgi/webkit-qt
