On Wednesday 16 January 2002 12:03 pm, Thomas Zander wrote:
> The more you make configurable the worse the usability gets of KDE as a
> whole. Please think before you make/suggest configurable. It is totally
> stupid to make this configurable in the first place!

Before more people get insulted let me give a short update. I seem to hav=
e=20
found a reasonable algorithm to determine the size of a messagebox. It wi=
ll=20
prefer a size larger than the minimum required size but smaller than 1/3r=
d of=20
the screen width. If the minimum required size is larger than 1/3rd of th=
e=20
screen width it will use that instead but it will never make the dialog w=
ider=20
than 2/3rd's of the screen.=20

This is still based on Qt's WordBreak routines. It might be possible to=20
squeeze words that don't fit into 2/3rd of the screen, but it will not be=
=20
possible to provide a tooltip for that word only. That is, the tooltip wi=
ll=20
be active for the entire text and I guess should either contain the=20
non-squeezed word or the total non-squeezed text.

If we want per-word tooltips someone should provide a widget that support=
s=20
that. An advantage would be that we could cache e.g. the size of words wh=
ich=20
would make the various operations required slightly more efficient. A=20
disadvantage would be that in order to stay consistent, it should also be=
=20
able to handle rich-text. Given that, it might be easier to hack Qt so th=
at=20
it squeezes words when they don't fit, (and provide a tooltip) than to=20
reinvent the wheel ourselves.=20

Since the consensus seems to be that automatic linebreaks are here to sta=
y,=20
that means that all mid-sentence line-breaks within message boxes should =
be=20
removed. Line-breaks should only be used to indicate end-of-paragraph.

Grepping the .pot (?) files for line-breaks which aren't preceeded by a '=
=2E'=20
might provide a good start.

Cheers,
Waldo
--=20
bastian@kde.org  |   SuSE Labs KDE Developer  |  bastian@suse.com

 
>> Visit http://mail.kde.org/mailman/listinfo/kde-devel#unsub to unsubscribe <<