On 23/02/12 16:37, Ryan Leavengood wrote:
On Thu, Feb 23, 2012 at 10:12 AM, Simon Taylor <simontaylor1@xxxxxxxxxxxx> wrote:Webkit also has a Cairo-based graphics backend along with the skia one. So perhaps a Cairo port makes sense for both a Firefox port and for maintainability of the WebKit port.I think getting Skia working in Haiku would be a better idea. As you said we have Google making sure the Skia backend in WebKit is good and fast, plus this is one more piece we would need for our own Chromium port some day. If Firefox Azure also had a Skia backend then even better: all the bases are covered.
Both Cairo and Skia have actively maintained WebKit backends, but Cairo is definitely preferred on the firefox side, which leads me to lean to Cairo for the initial investigation.
I'm planning on keeping the abstraction very low level - just wrap some memory as a surface, and not support concurrent BeAPI drawing access. There will be some investigation required as to the quickest way to get those surfaces into the app_server, but hopefully if there is any work required there it could be shared between Cairo and Skia and any other client-side rendering libraries.
You may want to do some searching though as aljen (Artur Wyszynski) did some work to port Skia to Haiku some time ago while investigating a Chrome port. I have an email from him with a link to his work-in-progress from 2009 though the link is now dead. I could see if I have the download on my Haiku machine though. Or maybe Artur is listening here and can provide any updated code for his Skia port?
That sounds interesting, I'd welcome news of any work that's already happened in this direction.