Nokia Engineers on KHTML

Rich writes "KDE could soon be making its way into your mobile phone. At aKademy in August David Carson and Deepika Chauhan from Nokia presented the work they've done in integrating KDE components into the latest version of the company's mobile phone software. Philip Rodrigues discusses this work with them on dot.kde.org."

Comments

Re:KHTML?

[biehl]

Yes, WebCore is Apples fork of KHTML. Read here for an explanation on how the collaboration between the projects works.

Re:Cool... BUT (there's always a BUT)

[TheRaven64]

WebCore is not dependent on X. It uses a small Qt-like backend. On OS X, this translates drawing calls into Quatrz (PDF-like) drawing commands. On X11, they can be translated into X11 drawing calls. On a Series 60 phone they will be translated into the native GUI's API.

WebCore is LGPL. They have to make their changes available to people who buy their 'phones (they also have to allow their customers to use a different version of WebCore, which could be interesting). They do not have to contribute their changes directly back to Apple, but they probably will since it is usually much easier to contribute patches than maintain a fork (note that this didn't apply to Apple when they forked KHTML, since they were throwing more developer time at the codebase than KHTML had in total, so it was easier to fork).

Re:Cool... BUT (there's always a BUT)

[RossyB]

2. What bothers me about an X system is that it is targetted at client-server, and the resultant code bloat may prove hazardous to an embedded implementation. I do however that an open-source-based solution should be used (why re-invent the wheel).

Sigh, this again. In X when the client and the server are on the same machine, communication is by local Unix sockets, which are the fastest form of IPC on Linux. Keith Packard wrote a new X server (kdrive) to demonstrate that X doesn't have to be slow, and he was right: the "overhead" of the client/server communication is nothing compared to the time it takes smaller systems to draw.

Re:Small RAM footprint

[Jessta]

it's not KDE. It's KHTML.
KHTML has a far lower footprint than something like GECKO(mozilla firefox).

Re:Cool... BUT (there's always a BUT)

[Bogtha]

Is this an appropriate GUI system to be using in such memory-deficient devices?

RTFA:

We started investigating the available open-source solutions and decided to go with a KDE-based solution for primarily 3 reasons:

1. Series 60 devices are constrained by ROM/RAM. WebCore/KHTML and JavaScriptCore/KJS provide rendering engines that use very small RAM/ROM footprint. That was really a clincher for us.

Re: KHTML

[g2devi]

There are two ports, one from Apple and one based off the work from Apple by Nokia. Here's the link I think you're referring to:
            http://gtk-webcore.sourceforge.net/

From the page: "Gtk+ WebCore is a Linux/Gtk+ port of Apple Computer Inc.'s WebCore KHTML html rendering engine including a web component. A reference browser implementation is included in the project. Gtk+ WebCore is a standards compliant (X)HTML rendering engine, javascript interpreter and an embeddable web component. The purpose of the web component is to be a light-weight, easy-to-compile and embed, open source rendering component.

The project work is done at Nokia Research Center (NRC) as part of ongoing internet browser-related research activities. By releasing the source we hope to support in open source communities interested in using KHTML rendering engine component."

Re:KHTML?

[fdobbie]

Since that was written, the world has moved on. Apple launched the WebKit open-source project as part of OpenDarwin. This means that WebKit bugs are now being tracked in bugzilla (in addition to Apple's internal bug tracking system), and WebKit, WebCore and JavaScriptCore have moved to a publicly accessible CVS server.

Re:Cool... BUT (there's always a BUT)

[rdieter]

WebCore is LGPL. They have to make their changes available to people who buy their 'phones

True.

(they also have to allow their customers to use a different version of WebCore, which could be interesting).

That's an interesting but (IMO) false interpretation of the LGPL.