[openbox] HCI-based suggestions for minor feature additions and behaviour updates

Clay Barnes clay at hci-matters.com
Fri Feb 1 19:16:08 EST 2008


On 13:28 Fri 01 Feb     , Dana Jansens wrote:
> On Fri, Feb 1, 2008 at 1:14 PM, Clay Barnes <clay at hci-matters.com> wrote:
> >  I'd be careful about moving the cursor---remember there are absolute
> >  positioning pointing devices (like the tablet I use) that generally do
> >  strange things when moved around, jumping back and forth and so on.
> >  Also, remember that you have to be very careful not to leave any
> >  changes in mouse position after operations finish---otherwise users
> >  will get frustrated quickly.
> 
> I've given some thought to the restoring the pointer thing too.
> Should it go back where it was, or should its final position change
> relative to how far you moved/resized the window?  That way it stays
> in the same position within the window more or less..

Hm.  I think that we should avoide moving the pointer at all.  That's
the expected case for users.  Though we might be able to enchance
certain situations by playing with its coordinates, I think that'll be
wholly negated by the frustration of the ones we don't predict making
decisions for the user without their control.  Though I personally
think that coordinate locking relative to the active window might be
the better design, I fear the world might not be ready yet... :-D

Also, the more I think about corner cases, the more I fear the risk
of bugs may be high.  What happens if the tablet tries to re-position
the mouse with absolute coordinates before the move completes?  What
about just when it's starting?  Ending?  What if the tablet tries to
move it, then releases it during the move?  During until after?  Lots
of bad things could happen, so if this were implemented, we should
definitely have it off by default.

> >  > Rather than this being a "while you hold down the modifier" thing, I
> >  > would prefer this to simple flash a little line on screen for a second
> >  > that is unrelated to modifiers.  (Keyboard grabbing should be avoided
> >  > at all costs in general..)  This could be really prettified by a
> >  > composite manager too. :)
> >
> >  Well, how is the current system handled?  I know that when I hold down
> >  the Meta4 key and use my vim keys to move around, the current window
> >  is stays selected until I release my Meta4 key.  That was the system I
> >  imagined expanding to cover this situation, too.
> 
> That's called an "interactive" action in openbox, and does require an
> active keyboard grab.  The focus cycling actions, and the "Move" and
> "Resize" actions are about the only interactive action, that I can
> think of right now.  The rest are just one-time "do this and done"
> type actions.  This type does not require any active keyboard
> grabbing, and is preferable as there is no issues with focus moving
> and applications not seeing it.  So, in other words.. the "MoveToEdge"
> type actions are not interactive and I don't want to make them so,
> they have nice behaviour the way they are.  So, having the bar flash
> rather than stay until you release a key lets their behaviour stay the
> same, and would work fine for the more interactive "Move" and "Resize"
> actions as well.

I suppose that a good compromise would be to do like we do with the
desktop switch dialog---have it appear briefly, and if another status
(in this case, a new edge/bar) becomes relevant, erase it instantly
for the new information.

> 
> dana
> 

One other thing---is anyone using the <middle>yes</middle> option
successfully?  I can't get it working here, and an example of a
working rc.xml with it might be a real help.

-- 
Clay Barnes

Website:
http://www.hci-matters.com

GPG Public Key (Fingerprint 0xbb14 fce2 1199 c413):
http://www.hci-matters.com/keys/claybarnes_public_key_until20080718.gpg
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://icculus.org/pipermail/openbox/attachments/20080201/926f233a/attachment.pgp>


More information about the openbox mailing list