Some questions.

[Davi]

Hello guys.

I would like to ask some simple questions about webkitdriver or more
precisely, on the headless webkit fork you made:

Is there a roadmap?
How difficult is implementing the render (I would like to help, if I
could)?
What's the path if one plan to use just the hl webkit portion?

Thanks in advance.

Best regards.

[Igor Viarheichyk]

Hi Davi,

Yes, there is a roadmap. In general there are two concurrent workflows, each have several tasks, in order of priority:

A. Headless platform improvement

1. Moving to most recent snapshot of WebKit. Current implementation is based on 1-year-old code, which is a bit outdated. However, there is a lot of code changes in WebKit (including code structure), so it requires quite a lot of time to sync up. Working on it now.

2. Rendering to PNG image on request. Planned, but not started. The main problem to be solved - rendering of fonts without introducing dependencies to windowing system (Linux). Probably Pango library is a best candidate to do this, but more analysis is needed.

3. Simple external interface. Current implementation implies only WebDriver interface, the plan is to make it a standalone command-line tool for basic automation tasks. The simplest approach is reading javascript from stdin, but need pay attention on multiple frames and virtual bowser window contexts.

4. Building for other platforms (MacOS-X, probably Windows). Currently it build for Linux only, building for MacOS-X may require minor changes, buidling for Windows looks like a huge task.

B. WebDriver API

1. Implementation of RemoteWebDriver API to make it interchangeable with any WebDriver without modification of source code of tests.

2. Implementation of recently added async javascript and similar WebDriver interface.

3. Moving to unified Browsr Atom infrastructure.

 

On Fri, May 27, 2011 at 1:32 AM, Davi <davirt...@gmail.com> wrote:

Hello guys.

I would like to ask some simple questions about webkitdriver or more
precisely, on the headless webkit fork you made:

Is there a roadmap?
How difficult is implementing the render (I would like to help, if I
could)?

I assume PNG/whatever image format non-interactive rendering. It requires quite a lot of effort because we can not rely on any windowing system. 

Thought basic rendering tasks are done by WebKit core itself, it requires implementation of several platform-specific bindings:

- Render UI controls (radio-, check- buttons, input fields, etc) WebCore/platform/hl. Currently we provide only empty stubs, need to implement drawing at least blank rectangles of proper sizes instead of actual controls.

- Render fonts (WebCore/platform/graphics/hl). Need a way to do it without dependencies to X-window (and minimizing dependencies on external libraries if possible)

- Rendering images. Jpeg and png images decoding already supported, just need to integrate with overall rendering.

- Rendering of image buffer to file. There is already dependency on libpng, so this part shoudl be relatively simple.

 

What's the path if one plan to use just the hl webkit portion?

For now WebDriver interface is implied. It is possible to implement a tool using WebDriver API which provides bindings depending on use case. E.g. command line tool which accepts URL as command line parameter, loads this URL, waits till it is loaded, makes a screenshot and puts it to file.

 

Thanks in advance.

Best regards.

--
You received this message because you are subscribed to the Google Groups "webkitdriver" group.
To post to this group, send email to webkit...@googlegroups.com.
To unsubscribe from this group, send email to webkitdriver...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/webkitdriver?hl=en.