On Sun, Jun 09, 2002 at 06:50:19PM +1000, Jeff Waugh wrote:
> > For things that fail to reopen on the previous workspace when you start
> > your session then yes we have badly behaved applications. But lots of
> > things get opened after the session starts, and matched windows can
> > still apply.
>
> - Do you need the full, outrageous functionality of matched windows to
> provide this feature?
I don't - the breadth of configurability found is remarkable, but totally
unencessary for practical usage. How to match the window will always be
the most hackish part I fear. While matching on class is technically
correct, some apps (Mozilla) don't set meaningful classes for all of their
various windows, leaving you unable to eg match just the browser or just
the MUA or ... whatever. While matching on class can be done cleanly,
matching on title requires either a complex set of matching rule filters,
or the ability to just specify a regex. I hate both, but the latter is
most approgriate in a UNIX environment, and the least hackish of the two
options.
Once you match the window, Sawfish gives you a million choices for what to
do with it once matched. No way most people need all of that crap. All I
normally use are workspace placement and whether the window should be
zoomed on mapping or whatever. I also have matches which use the never
focus and completely ignore it flags, mostly for ill-behaved special case
type things. Also, panel and nautilus have their depth fixed because of
... some app I used to use (forget now) which broke things in that regard,
it's probably safe to remove that now.
> - Do you think a simpler, more obvious interface to this problem could be
> created?
Sure it could be better. Radio button for matching on class (this one
kind of window, or all windows from this application) or title (enter
regex here). Option to select a default workspace (Default: current) and
whether to zoom/iconize the window (Default: do neither), and a couple of
default-off checkboxes under the "advanced" heading, including the weird
options the advanced user should feel comfortable knowing he need never
alter such as focus, sticky, ignore the window, etc.
Drawing a mental dialog (too tired to fire up glade and draw it on screen)
I don't like how it comes out - but it's better than what Sawfish has now.
Of course, in the context of Metacity, it's all up to Havoc. If this was
just a thought experiment, it confirms what Havoc was saying durring my
last discussion with him on the subject: it will always be a hack. Since
I spent half that discussion agreeing with him, I doubt I was successful
at convincing him it was important, hack or otherwise.
OTOH, if you're seriously interested in simplifying the madness found in
Sawfish and think there's a chance Havoc may be convinced to do something
useful in that area, I'll take a few minutes to put together something I
think is reasonable tomorrow morning. Might not come out like I describe
above though since once I see what I've described in dialog form I'll
undoubtedly decide it's quite wrong and redo it. =)
> When I used Sawfish, I solved similar problems with the matched windows
> feature. My favourite was setting GIMP to be click to focus, whilst my
> default was sloppy focus. This solved my problems with lots of toolboxes,
> and a very active mouse pointer (given how I work with The GIMP). I also
> used to set my GnomeICU window to use a different, less-chunky WM theme.
I'd never considered sloppy/click focus to be a useful thing to set. Now
that you mention it, I can think of ways it'd be useful. (I normally use
click to focus myself..)
> I use metacity now, and I miss these tweaks! (Kind of... I don't use any IM
> software at all anymore, and I can always set metacity to use click to focus
> when I'm doing lots of GIMPing.)
>
> Anyway, so yeah, I fully agree that the matched windows feature could do a
> lot of cool stuff, and was often quite useful. I used it a lot myself.
>
> However, many GNOME hackers are no longer content with hundreds of options
> that essentially fix user interface breakage, and certainly no longer
> content with overblown features that don't solve user problems in a direct
> fashion.
Understandable. Generally I agree, for most minor features I'd rather
have a sane choice which can be applied to everyone over poor choices
which can be changed. Of course I object to poor choices which can't be
changed, but these are sometimes highly subjective. I try not to get too
annoyed unless I've got real justification for why it needs to work
differently.
> Many of the things we use matched windows for can be fixed better, and for
> more people, with some creativity and considered thought about how we work
> with our software.
Granted. My own example above about ignoring a window: Most apps which
have windows which should be ignored tell the WM to ignore them with the
correct hint. Those which don't, should. The weird exception cases like
the syslog tail are interesting though - I don't have a good solution for
them, and am undecided whether those should be considered or not.
> Much like earlier debates about desktop navigation, it comes down to
> defending *behaviour*, not *features*. Those are the kinds of bugs and
> changes that we're really exicted about right now.
I agree with this model, actually. If you know what you want to
accomplish exactly, it is much easier to find all the necessary bits to
make that happen. If the task cannot be solved adequately with what
you've got, that's about the time when you start thinking about adding a
missing feature somewhere. Looking at it any other way leads to feeping
creaturism and lots of bloat.
The sig picker is scaring me tonight.
--
Joseph Carter <knghtbrd@bluecherry.net> Sanity is counterproductive
<xtifr> Athena Desktop Environment! In your hearts, you *know* it's the
right choice! :)
* Knghtbrd THWAPS xtifr
Attachment:
pgpzjtyEAi1_q.pgp
Description: PGP signature