Tristan:
The main problem I think with that model, is that you will still have programs designed for one specific platform, ie. the desktop; and then have the UI essentially jury rigged to other platforms, using various UI plugins, resulting in a substandard UIs on the other platforms.
It's probably true that there would be bugs. Today some applications using the ATK do not work well for blind users because the designers didn't take certain a11y usability issues into consideration. For example, how focus order affects a blind users understanding of the overall interface). I'd imagine if an ATK-like interface were used to make GNOME work on different types of devices, there'd be similar bugs that would need attention over time.
Programs running on iPhones for example are constrained by screen space and use very different input devices to their Desktop counterpart. It would then make sense to write completely different UIs for the two platforms, to take advantage of the features of each. A complete separation would for example allow Canonical to really customise their flavours of Ubuntu for different platforms.
The ATK interface currently supports users who are blind and users who are not blind using very different usability models. If the existing interface can support such wildly different models, then I do not see why we would assume that any non-desktop UI would be substandard. I'd say most existing GNOME blind users would, in fact, say otherwise. Text-to-speech, in general, works better on a wider set of applications in GNOME than it does on Windows, for example. That said, the ATK has never been used to try to represent two different graphical user interfaces in an interchangeable fashion. It would probably require some not insignificant effort to make it support this sort of thing. But probably considerably less work than designing something completely new from scratch. The main thing that is missing from the ATK is any sort of graphical hinting to help a widget set determine how things should be visually arranged. This is not as much of a concern in the blind-user case as it would be for creating a graphical interface like an iPhone interface.
You would also not recieve many of the other benefits I listed in my previous email, such as more UI experimentation.
That is very subjective, and many is a fuzzy word. Try being more specific, and it would be easier to dig into these topics, I think. Brian _______________________________________________ desktop-devel-list mailing list [email protected] http://mail.gnome.org/mailman/listinfo/desktop-devel-list
