Re: mobile-devel-list Digest, Vol 17, Issue 23

[Bob Murphy <bobert lavitanuova com>]

Stormy wrote:
> If we are using existing GNOME desktop technologies and we aren't  
> optimizing
> those technologies for GNOME Mobile, then what are we doing? Are we  
> writing
> connectors, sharing experiences, are we a users group? A reference  
> platform?
>
> I'm ok being cement (sorry, Dave ;) but I guess I'm asking are we  
> rebranding
> GNOME cement as GNOME Mobile cement? Or do we actually do anything  
> to it in
> the process? Like change it or integrate it with other types of  
> materials in
> a way that's good for mobile? (Either one is ok for me, I just want  
> to be
> able to articulate it cleary.)

We are, in fact, optimizing GNOME technologies for mobile use. It's  
just that such changes, so far, have also been of benefit to desktop  
users, or at least not detrimental.

For instance, Nokia's team did a yeoman job of replacing floating- 
point calculations in Cairo with fixed-point math. That resulted in  
some speedup on desktops, but massive, fantastic acceleration on ARM  
chips, which typically have to do floating-point math in software.

Similarly, I've made some changes to GDK and Pango (yup, need to  
submit those... :-) to fix a set of mobile-specific bugs related to  
non-atomic queries to X. You'd never see these on a desktop (well, not  
without getting a hernia), but if you take a small portable device  
that supports Xrandr and has orientation detection, and rapidly flip  
it between landscape and portrait orientation, weird things eventually  
happen to window and font sizes.

There are other areas that could use mobile-specific work. One that  
comes immediately to mind is that certain GTK widgets, such as  
GtkColorSelectionDialog and GtkFileChooserDialog, are much too large  
to fit on a cell phone screen. It would certainly be of benefit to the  
mobile development community to have alternate implementations or  
layouts for these widgets that can be used on small screens.

Let's say there's a spectrum. On the left is, "we're rebranding GNOME  
desktop but not really changing it." On the right is what's happened  
with OpenGL and OpenGL ES, which are related implementations for  
desktop and embedded systems, but aren't code-compatible.

As someone developing for GNOME Mobile, and watching what other people  
are doing, I'd say we tend toward the left - but we're not all the way  
there. It seems to me we clearly want to maintain a single code base  
for all GNOME technologies, but those of us working in the mobile  
space have no compunctions about making mobile-specific enhancements  
to that code base, as long as they don't damage other GNOME usage.