[rekonq] Re: [RFC] Better Integration of Web Applications into Window Management

[Martin Gräßlin]

On Tuesday 11 January 2011 20:05:17 you wrote:
> On 01/10/2011 07:11 PM, ext Martin Gräßlin wrote:
> > My idea would be more revolutionary and is nevertheless technically
> > feasible. Rekonq would have to tell us  (kwin/plasma) which web
> > applications are available (either through dbus or X), there names,
> > their icon and their thumbnail. Now KWin could include these web
> > applications as "children" to the actual window. Both our effect
> > framework as our alt+tab solution are abstracted enough from windows to
> > allow this. E.g. alt+tab is also able to manage desktops. Also in Plasma
> > I think that this is no problem to integrate as there is the
> > libtaskmanager and the tasks dataengine.
> 
> The more I read about this, the less I understand what is the
> revolutionary idea. :(
> 
> Could you summarize that in an example? Let say I have a web app:
> "Google docs". It registers to the browser to handle .odt files and be
> available offline.
> 
> What do you suggest on top of that?
> 
> My point of view is the engine should add an icon for such apps (the
> user is asked for that of course), and the icon opens a chromeless
> browser window.
> If the same app is opened in the browser, it is not chromeless, it
> behave as usual window.
My idea is not to have "new windows for webapps".

My idea is like the following:
You have one browser window with multiple tabs:
* GMail overview
* GMail write new mail
* Facebook page 1
* Facebook page 2
* Google Docs document 1
* Google Docs document 2
* 40 additional tabs with random browsing stuff

What rekonq would do now is to add markers for GMail, Facebook and Google 
docs, so that they are known to the Desktop Shell. The 40 random tabs are not 
influenced. To decide what is marked for integration into the shell the 
"Pinned Tabs" could be used (so it does not have to be a web-app).

This is quite orthogonal to the "one window per web-application". It exposes 
selected tabs to the desktop shell. It's nothing to replace the one window 
approach (which I would like to have, but does not exist yet, so it's mood 
point to discuss it) and even if the one window mode would exists it would not 
influence this feature at all.

So how do I think could this work e.g. in Alt+Tab. Alt+Tab uses a Model. This 
could be changed to a tree model, so that the exposed tabs could be shown 
together with rekonq.

The idea is mostly motivated by studying users who try to alt+tab from a 
google result page to the web application running in the same browser.

Just to point something else out: One window approach is quite difficult to 
get it right and if not got right it is annoying as it spawns the workspace 
with windows. In opposite to that exposing pinned tabs to the workspace could 
be a rather trivial feature to integrate into rekonq. The more difficult part 
would be to extend the workspace. As I understood you already have a pinned 
tab feature coming. All that is needed is setting appropriate X hints and 
react on a change request.  Everything else would have to be added to the 
desktop shell.

Furthermore this could be implemented as an experiment without changing the 
workflow of users (compare Plasma Activities).

Cheers
Martin
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 316 bytes
Desc: This is a digitally signed message part.
Url : http://mail.kde.org/pipermail/rekonq/attachments/20110111/b70884ae/attachment-0001.sig