[RFC] New (QML) Desktop Containment

[Marco Martin]

On Thursday 22 November 2012, Aaron J. Seigo wrote:
> 
> * granularity of and visibility duing dragging with a finger is pretty poor
> * given the small screen real-estate and the effort needed to move things
> around, moving things to the nearest open space is pretty common
> 
> on desktop, i'd suggest trying to always drop things where they are placed
> by the user and never move it elsewhere for them. that may mean resizing
> other widgets.

so, to summarize the issues:

1, positioning)
* should be possible to put a plasmoid in any position
* but at the same time should also be easy to do a perfect alignment

forgetting for a while about the implementation, what would probably work the 
best from an ui point of view is (not too unlike the current behavior, but a 
bit fancier):
* plasmoids freely movable, if some conditions don't happen, they stay exactly 
in the pixel they are dropped.
* the drop target doesn't show normally, it appears when an applet would be 
then moved or resized by the layout system (ie user is dragging one on top of 
another)
* when put near another one, plasmoids snap at edges/corners to perfectly 
align: the drop target appears also when a snap would occur

now this is probably a bit computationally complex to be executed perfectly 
(especially doing that in javascript, ouch :/)

2, handles)
* concerns of applets starting to look too much like mini windows could have a 
point
* tough i also feel current handles to feel a bit wrong/on the way. Could be 
too much stuff happening when hovering an applet with a mouse, the handle 
appearing animation is something that screams look at me, instead of being 
something discreet that is usually not used (except the 0.1% of the times one 
goes with the mouse over for resizing)
that is probably the sense of clutter right now, so the question is, how make 
it more discreet?

3, about lock)
* as already linked, modes are historically been found to not work very well 
in tests
http://www.usabilityfirst.com/glossary/mode-error/
(forces the mind to do an extra error prone context switch that same times may 
be avoided, iirc there was even a famous plane crash which most probable cause 
was a misinterpretation of a mode in the cockpit instrumentation)

* however is true that many prefer plasma in "locked mode": why? I think it's 
a perception of less clutter and more reliability there is. So one thing again 
the applet handles that are right now perceived as annoying, two, accidental 
applet move/delete, so the problem is actually those two things:

* visual appearance
* accidental action trigger

can those be solved, at least in part without a mode to disable such 
functionality? (or at least make that mode not perceived as necessary)

Cheers,
Marco Martin