k16

[Chani]

well, this isn't going to be much good because I'm not able to believe 
anything good will ever happen at the moment, but since the deadline is coming 
I might as well send *something*.


what's been important to me lately is context and sessions.
we've already got plans that in 2011 we'd get context-awareness in apps. 
they'd start responding to the activity they're in. and rrix wants to add 
location context stuff. you already know about that, though.

what you may not have thought of is the opportunity that wayland presents: if 
we can get a new session protocol in - a spec that covers both the management 
and the individual apps' session storage - we can create a system where 
activities are *mobile*. Where you can just transfer your current activity to 
your n900 and walk out the door, taking your documents with you - even if you 
don't have the same applications on both devices (eg. koffice on your laptop, 
freoffice on your phone - they may be from the same project but they're not 
the same app :).

KDE could be truly mobile - not just in the sense of having mobile versions of 
our apps, but having the the user's actual data mobile across devices. We can 
transcend applications. ;) A device running KDE software becomes a context-
sensitive, adaptive environment, helping the user without getting in their 
way, and sharing freely with their other devices (so long as those devices are 
standard-compilant.. *cough*meego*cough*)

imagine having some research open, and being able to bring it along and keep 
reading on the bus. or having looked up directions to a restaurant and 
theater, tossing those onto your phone so that you don't have to re-google if 
you need to check again. or maybe you're halfway through your favourite tv 
show when you get dragged out the door - but you can take your netbook and 
keep watching it instead of waiting until you get home. :)

how do we do this? we convince apps to store their session data in nepomuk, in 
a standard way, so that similar applications can share the data. The session 
*itself* becomes document-oriented instead of application-oriented. Even if 
one device has a pdf viewer with tabs and the other can only handle one pdf 
per process, the session can be transferred losslessly if it's saved in the 
right way. :)

the biggest problem, of course, is convincing apps to port to this. the 
nepomuk guys had thought about this before, it seems, but they don't think we 
can convince apps to switch. I think we *can*. :) and wayland  will help. if 
wayland takes off, if it gets popular and apps start porting to it, we can 
make this *part* of that transition. We don't need every single app in the 
universe to join in, just enough for the idea to gain traction - and if kde 
apps are first, they'll have an advantage. :)

wayland doesn't have any session spec at all yet. nobody's thought about it. 
not only that - gnome hates the XSMP session protocol and wants it to die. I 
can give them that, and give us something really fucking awesome. :)

I don't know how long this'll take; I haven't looked at the wayland timeline. 
I know it won't be quick, though. it's going to take a few years (hopefully 
not the whole five!). Obviously there'll be legacy xsmp support for a long 
time. One thing that'll help: we can begin KDE's transition to the new session 
storage before kwin-wayland is implemented, since XSMP doesn't specify 
anything about how apps store anything.

I'm also thinking that with it being document-centric, when the server side 
*is* implemented it could take more direct control of what gets started how - 
for instance, on a phone, where there's not enough resources to open a dozen 
documents all at once, it could display a list of resources and not restore 
them until they're clicked. Or open just a few.

During syncing, it could also skip Very Large files (you're not going to try 
editing a movie on your phone). Oh, and it could also be "abused" for app 
introspection (eg. the window manager knowing what tabs are open - although 
that particular feature may be in 4.7 anyways). There's lots of fun details 
here. :) Lots of opportunities for people to expand on it with their own 
ideas...

...err, crap. this might be a bit long for one page, eh? :) can I write a 
cleaned-up version later when I'm in a better mood? :) I haven't even talked 
about context... 

-- 
Chani
http://chani.ca
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part.
Url : http://mail.kde.org/pipermail/k16/attachments/20101228/15906c92/attachment-0001.sig