[Desktop 13.04-Topic] Having a smart ubuntu desktop

[Didier Roche]

Le 18/10/2012 17:14, Jacob Williams a écrit :
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On 10/18/2012 11:10 AM, Didier Roche wrote:
>> Hey everyone,
>>
>> In a world where we are using more and more connected web services
>> doing some of our tasks (web mails, online documentation editing,
>> online music players…) should we imagine having a more adapated
>> image to our users? This will mean reducing our main image
>> footprint by removing some of those tools we install by default:
>> I'm thinking of thunderbird, libreoffice, rhythmbox and other main
>> applications of our desktop for instance.
>>
>> The counter-part would be to make our desktop smarter. I can
>> imagine: - having the messaging menu (or an icon in the launcher,
>> or an icon in the dash) showing, the first time you try to
>> configure your email account, a window asking for your email -
>> based on the answer, either proposing to directly use a web
>> application (with unity integration) for an @gmail.com, @yahoo.com…
>> and other email providers known to have good web integrations.
>> Otherwise, proposing to install thunderbird, ideally opening the
>> account creating setup prefiled with the information already be
>> done. (we can of course imagine a checkbox to override the "smart"
>> behavior).
>>
>> This is just a field example, we can expand to document editing,
>> and a lot of other areas. There is already some integration of this
>> for other parts of the stack (like double clicking on an odt when
>> you don't have libreoffice installed), we can make sure all our
>> desktop have this kind of tweaks, and try to make a sharper, more
>> adaptive image to our finale users, without having lots of
>> post-install applications to remove.
>>
>> What do you think? Is that a goal worth having? Cheers, Didier
>>
> Hi,
>
> It seems that you're proposing removing applications that
> differentiate Ubuntu from other platforms without a providing a reason
> why this wouldn't be catastrophic for users.
>
> This would decrease the functionality available to users. Web
> applications are available from the browser without introducing extra
> dialogues and components which may confuse users or fail, or changing
> the distribution significantly by removing a large number of
> application already in use.
It's not "already in use" for new installs. Of course, those kinds of 
functionalities only impact new installations.
>
> The browser is the interface targeted by web application developers,
> and currently works with all the web applications that you've cited.
>
> I can't see a reason to presume that these applications could be used
> from a assembly of different programs without changing the experience
> users are accustomed. This would rely on many components which would
> have to be developed then tested and supported for the life cycle of
> the web application current user experience and protocols, which are
> subject to change at any time.

I think those are valid reasons here, however, if we try to target the 
main user audience, a lot of people doesn't really understand the remote 
versus local paradigm, and that you can try to fetch remote data and 
consult them locally. For instance, webapps like gmail, free, yahoo: I 
see the wider computer users not understanding that they can read their 
email locally and what is on the online website user interface is 
exactly the same, but just a different view from the same data (maybe 
pop3 with no "keep on server" is to blame here). The result is that they 
don't use those kinds of tools (it can be interesting to see how many 
people configured thunderbird on their machine).

So why installing them at first? Why not providing a great out of the 
box experience for them like "I click on this email icon and it 
connected me to the corresponding webapp"?

>
> Assuming all targeted web applications maintain the same user
> experience and protocols for the life cycle of an Ubuntu release,
> making the current components of GNOME such as the launcher able to
> use web applications with local or remote files would be a huge task.
It's already the case in 12.10, look for the webapps feature we 
introduced during the quantal release cycle development :)

>
> Consider your case of opening an ODT with Google Docs, gnome-vfs would
> have upload the document to a Google account through whichever method
> of doing this works right now, then the launcher would have to detect
> if this succeeded and create a window with only the Docs interface
> rendered and then the window manager would have to track all new
> windows spawned from this 'top level' window as being part of the same
> window group (how would BAMF handle this?) for the Unity launcher, etc.
Downloading/uploading documents is more complicated than the email case 
(and that's why I think we should start with email first), the matching 
and window/tab issue is already working in bamf today. That's why I 
think we should focus on this documentation transmission if we want to 
do it (at all) during the UDS session.

>
> While I can imagine a desktop that did all these things, while
> maintaining the current set of tools and not requiring a behavioral
> change for user would be very cool .. I think your idea of how this
> would work is not good because it would remove nearly all current
> functionality and be architecturally complex.

I don't really agree: we don't remove anything, we are proposing a smart 
way of detecting the user intent and habits, to make the desktop more 
personal and adapted to the user itself. All the apps we already shipped 
would be in main, still maintained by the desktop team. So no change in 
any case for people liking using libreoffice for instance.

> Please do not remove all existing functionality chasing web
> application nirvana
>
> As an aside, Desktop Team .. why is not possible to add more
> functionality instead of replacing current functionalities with a new
> and non overlapping set of new and untested functionalities? Why does
> each new paradigm introduce SERIOUS (I cannot overstate this)
> regressions over the last paradigm.
>
> Ubuntu no longer runs in a virtual machine because of it's 3D graphics
> requirements, in 2006 this would have been considered a Very Bad
> Thing, why is it acceptable now?

This is a separate topic covered already, but feel free to create a new 
thread (you would notice 2006 is 6 years after 2012 and that hardware 
evolves…) :)

Thanks,
Didier