[haiku-development] Re: R1/a4 initial planning

Den 22 februari 2012 18:30 skrev Stephan Aßmus <superstippi@xxxxxx>:
> Am 22.02.2012 16:17, schrieb Ingo Weinhold:
>> But even if (re-)porting Firefox is not considerably less work than
>> catching
>> up with WebKit, it would at least bring us a complete browser.
>
>
> I don't think this is or has ever been much of a "complete, full-featured
> browser" versus "small browser with only the essential stuff" problem. This
> has always been a problem of reliability and integration.
>
> BeZilla managed to get more and more reliable. Only the latest versions got
> reliable enough. In the intergation department it never got far enough IMHO.
> The latest version still prevents a regular system shutdown.
>
> Similarily, the problem with WebPositive is not that it has too few features
> compared to a "full browser". I think the problem is that it is not
> reliable. Both in terms of randomly failing in the same situation that it
> worked in before, as well as not being usable for many situations at all
> (think of accepting an untrusted SSL certificate). Then the port itself is
> incomplete, for example transformations are not supported in the rendering
> layer, the problems of the network layer, no caching.
>
> I think if the WebKit port would be more complete (most importantly
> rendering and network backend), and if it would "just work" and be 100%
> reliable, then almost nobody would complain it can't do this or that feature
> that Firefox or Chrome can do. A good, native integration would almost
> certainly feel more important to a lot of people.

I havn't looked at recent Firefox, but it may or may not be easier to
port. The trend towards the end seemed to be to make it harder to
port, instead of fixing problems in NSPR they wrote something that was
almost the same in XPCom but less portable. Add to that the need of
new GUI and font-handling and probably a lot of upgraded or new third
party libraries they will have added to that.

Sergei Dolgov and I worked a lot to keep up with other platforms, and
it was a lot of work. Since WebKit seems to be a much faster moving
target and with a less clear roadmap it is probably much more work to
maintain.

What Stephan mentions as problem are actually quite easy tasks to
solve. The shutdown problem has been fixed (it's caused by the
horrible internal relaunching) and the only thing left for proper
intergration was negotiation between firefox object mimetypes and BeOS
mimetypes. (I never could find motivation for that one).
What drove me of from continuing working on Firefox was the negative
energy that you always got from other devs, so for me I'd pick WebKit
any day of the week.

Btw Ryan, I'm still interested in your Webkit repo. It would be nice
to have a common one.

/Fredrik Holmqvist, TQH

Other related posts: