[prev in list] [next in list] [prev in thread] [next in thread] 

List:       webkit-dev
Subject:    Re: [webkit-dev] Changes in QtWebKit development
From:       Konstantin Tokarev <annulen () yandex ! ru>
Date:       2013-09-30 13:48:43
Message-ID: 360351380548923 () web12j ! yandex ! ru
[Download RAW message or body]



30.09.2013, 17:39, "Dirk Schulze" <krit@webkit.org>:
> On Sep 30, 2013, at 11:58 AM, Allan Sandfeld Jensen <kde@carewolf.com> wrote:
> 
> > On Thursday 26 September 2013, Andreas Kling wrote:
> > > On Sep 25, 2013, at 12:40 PM, Allan Sandfeld Jensen <kde@carewolf.com>
> > wrote:
> > > > On Saturday 14 September 2013, Andreas Kling wrote:
> > > > > On Sep 14, 2013, at 11:24 AM, Allan Sandfeld Jensen <kde@carewolf.com>
> > > > wrote:
> > > > > > That said, in all likelihood the Qt port will not remain part of WebKit
> > > > > > forever, ...
> > > > > (This being the main reason.)
> > > > > 
> > > > > Since you already know you’re eventually going to leave, you could just
> > > > > move to a branch sooner rather than later. It’s unreasonable to expect
> > > > > WebKit to accommodate a port that has no forward-looking interest in the
> > > > > project.
> > > > We do have a  branch tagged and being prepared for 5.2. It was taken
> > > > before the FTL merge and the following switch to require C++11 in all of
> > > > the project. It will be very hard branch again after that point since we
> > > > support 2-3 year old platforms by default, and the Webkit project want
> > > > to move to using the latest and greatest compilers.
> > > So you are saying that you'll never branch QtWebKit from WebKit trunk
> > > again?
> > I would love to, but I do not think it is going to happen. Quite honestly I
> > wasn't sure I would be able to pull a new branch for 5.2 off, since older
> > Linux (gcc 4.4), all windows builds and especially old OS X (10.6) were not
> > building WebKit2 when I started. I got it working, but it the work to unroll
> > unnecessary compiler features and library dependencies is just going to get
> > harder from now on (if anyone want a patch to remove the C++11 requirement
> > from WebKit2 late July, I have one).  If a new branch is made from WebKit
> > trunk in the future would likely only be limited to specific platforms, and
> > therefore not suited as a module shipped with Qt, but as an optional upgrade.
> > > It’s commendable that you want to land your platform-agnostic patches
> > > before withdrawing from the project, but assuming your last branch point
> > > is already set, I don’t see why this necessitates keeping the Qt platform
> > > code around.
> > We all know what happens when a webkit port works on a branch. In theory it
> > shouldn't be a problem, but as you know it didn't work for the N9 browser
> > branch in Nokia, it didn't even work for the iOS branch at Apple!
> > 
> > So based on observations, I believe to be part of the project and able to
> > commit upstream you must live upstream.
> 
> I would not necessarily disagree with the problem of upstreaming work. But you said \
> that most likely you wouldn't be able to branch WebKit anymore because of the \
> compiler requirement.  At least for Qt. Do you have other interests in QtWebKit \
> beside the integral part of Qt so that it makes sense for you to maintain the port \
> further? 
> Another question that is just partly related to WebKit but more curiosity.  Qt is \
> deep integration into WebKit. We have (had?) a lot of Qt specific code in core \
> WebCore to support QtXML and other things. Blink already stated that they would not \
> accept such deep interventions in their platform. Is all that not important for you \
> anymore? Can you operate with libxml2 and other libraries from now on? If that is \
> the case, can't we limit the Qt specific code to just /platform/qt and remove all \
> other Qt specific dependencies from WebCore?

For embedded targets libxml2 would be an additional dependency to build, and it would \
take additional space in device firmware. New XML parser from [1] would be better \
alternative if it was finished (and had decent performance).

[1] https://bugs.webkit.org/show_bug.cgi?id=64396

-- 
Regards,
Konstantin
_______________________________________________
webkit-dev mailing list
webkit-dev@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-dev


[prev in list] [next in list] [prev in thread] [next in thread]