[webkit-dev] Regarding OpenVG

[Jakob Petsovits]

Hi Madadi, Mustafizur,

Things have not been going as planned with the OpenVG upstreaming effort, 
partially because of my lack of drive to work on it in my free time, partially 
because of internal priorities that made it harder for me to find time during 
work to finish up the initial upstreaming that we started last year.

The current state of OpenVG in WebKit trunk is incomplete, in addition to 
having no buildsystem, it also lacks a few parts in the code that would 
otherwise be necessary to build an OpenVG-based WebKit. I opened up a new bug 
for every new commit that we added, one by one, so the bug that's still open 
is the last piece that I got around to put into the queue, however it's not 
the last piece that needs to go in.

A complete set of patches that I had at that time, that I was trying to push 
upstream, can be found at [1]. However, that's based on WebKit from March last 
year, i.e. ancient, and misses out on a many changes (fixes, API updates) that 
we've been doing afterwards.

(For reference, the commits in there with a commit message in the first line 
were not going to go into review, they were just there to modify a QtWebKit 
desktop Linux build to work with OpenVG. Also the font backend is not meant 
for productive use as it hardcodes a single font and leaks its glyph paths, 
let alone do anything useful in terms of advanced layouting. On the 
BlackBerry, we used a different font engine based on system APIs. Despite 
being old and lacking features compared to our current internal version, it 
should be possible to build the branch on Gitorious as is.)

So that's the bad news. Good news is that I'm allocating time in the next 
short while to work on cleaning up, squashing & rebasing of commits, and 
upstreaming of this work. Mind that there are no plans here to implement a 
proper non-BlackBerry font backend for OpenVG, anybody interested will have to 
do that by themselves.

As for the OpenVG reference implementation, we used to work with a modified 
version of that (implementing its mini EGL on top of Qt) but Matteo Muratari 
and Mazatech were kind enough to provide a pre-release, developer version of 
their AmanithVG for Linux (see [3]) which still exists at that link and, I 
believe, works out of the box with the Gitorious branch and is magnitudes 
faster than the reference implementation.

Furthermore, the released WebKit-for-BlackBerry code can be found at [2], and 
contains a current version of the OpenVG code. It doesn't come with commit 
history, but if you can live with that, you'll get the whole shebang in one 
place. Make that work with the font backend from my Gitorious branch and 
integrate it into your version of WebKit, and you're golden.

I'm usually on #webkit (as jpetso) to answer questions, feel free to contact 
me there although I might not have a good answer to everything.

Hope that helps,
  Jakob

[1] https://gitorious.org/~jpetso/webkit/jpetso-webkit/commits/openvg_9
[2] http://us.blackberry.com/apps-
software/blackberry6/browser_open_source_components.jsp
[3] http://marc.info/?l=webkit-dev&m=127117284405700

On July 17, 2011, Mustafizur Rahaman wrote:
> Hi All,
> 
> Some quick reference to the analysis we have done so far:
> 
>    - The master bug for OpenVG work was
>    https://bugs.webkit.org/show_bug.cgi?id=33987 where except one, all the
>    dependent issues were resolved, so we assumed the basic functionality is
> up and running...
>    - We also found https://bugs.webkit.org/show_bug.cgi?id=47475 while
>    scrubbing bugilla for openvg related changes, where someone commented (
>    https://bugs.webkit.org/show_bug.cgi?id=47475#c4 ) that the OpenVG
>    upstreaming effort might have been abandoned, though we did not get this
>    confirmation from anywhere else yet.
>    - Also as per
>    https://lists.webkit.org/pipermail/webkit-dev/2010-January/011253.html
> the backend functionality seems to be working except the build system is
> not yet public, but this was an older email anyhow.
>    - So, as Smitha mentioned below, we tried to use the sample openvg
>    implementation from Khronos (
>    http://www.khronos.org/registry/vg/ri/openvg-1.1-ri.zip) & tried to
> build the WebKit with OpenVG enabled & came across various compilation
> issues.
> 
> Therefore, if any one can throw some light on the questions below..
> 
>    - Is the OpenVG upstreaming effort still in progress? What is the
> current status?
>    - Any help as how to build the OpenVG code? If any one can help us, we
>    can go further details about the issues we are facing either in #webkit
> or in the mailing list.
> 
> Thanks for the help in advance,
> Regs,
> Rahaman
> 
> On Fri, Jul 15, 2011 at 5:54 PM, smitha g <madadismitha at gmail.com> wrote:
> > Hi,
> > 
> >       I am trying to build the Webkit code on Windows with OpenVG
> >       enabled.
> > 
> > I have used the openvg header files and lib of Kronos group but the build
> > fails. Could you please let me know if OpenVG code(under the macro
> > PLATFORM(OPENVG)) upstreamed is functional. If so, could you please let
> > me know the build procedure with OpenVG enabled. Also, kindly let me
> > know if the upstreaming efforts on OpenVG front have been discontinued.
> > 
> > 
> > Thanks & Regards,
> > Madadi Smitha