It all depends on toolsets available.
Eg, if Meego (Nokia) ever ships, then python + pyside + pyqt
means we can reuse large parts of gramps, but must redo the UI
front, which is needed anyway for the small form factor.
Probably need an sql backend too, as our database backend will
likely not be supported, but that is doable.
For the iphone I see there is a python kit, http://stackoverflow.com/questions/43315/can-i-write-native-iphone-apps-using-python
but it is not really supported. So you would need developers
that can code in Obj-C. Apart from that, you can reuse the xml
files and the datamodel, but the rest you must do from the
ground up. Not very interesting for the gramps developers.
Android has the same problem. Better to root the device, and
then install full linux.
Alternative is writing for the cloud, which is being started
with the gramps-connect approach. So a django server, and all
webbrowsers can update your data.
So, from this, Gramps on the desktop is here to stay, and
let's hope the toolkit we use get's decent touch support,
although we all know our present application and touch is not
the best marriage.
The cloud is an approach followed, but still needs quite some
work, help wanted.
If Meego ships, I'm interested in trying some QML to write a
family tree browser for that platform.