@qwerty12:

How well do the bindings match the repository webkitgtk API?

The webkitgtk in there is Bundyo's latest (well, minus the localstorage patch) and these were built against that so, on that front, all should be well.

As for the bindings, codegen says the methods have 100% coverage so that should be OK too.

I hope I didn't misread your question

I saw where you were talking about midori in the Tear thread. I got mildly excited 'cause those guys are using this gjs stuff, and I was hoping you had created the python bindings for it. The guy that created and maintains the pywebkitgtk project yanked his gjs bindings; I don't think he ever got them to work, and I couldn't even get them to compile with the way he had things set up.

Anyway, IIRC, this gjs stuff allows access to the DOM inside webkit. Have you ever used Safari on an iPhone or an iPod touch? Well, the iPhone OS webkit has this cool feature that I think is called "Web Spot": By double tapping, the view will zoom or unzoom based on the inclosing tap's <div> width. It's f*cking awsome. But I think the only way to do such a thing is to have access to the DOM. Blah, blah, blah...

So, the reason I asked was my impression from the pywebkitgtk code is that some of the calls need an override file, and I was just curious how deep in the sh*t you were. But if codegen is handling everything, cool. Here's some links just for the heck of it:

pywebkitgtk source tree
midori's gjs.h

May The Force be with you. Always.

And you need different zoom technique.

You're right, damn. It's just a view transform call. Except for rotation and view selection, there's really no resizing of the browser and there isn't any reformatting really, which means they're not even using webkit's zoom functions. And since everything is sitting on top of Cairo, I wonder if it wouldn't be just one matrix transform call in GTK. Those TI graphics drivers would sure come in handy right about now.