Re: Re: [dwm] dwm-1.7.1

From: David Tweed <>
Date: Thu, 28 Sep 2006 17:43:54 +0000 (GMT)

> Note that this is just pointing out that with multihead the
> "human issues" are at least as important as extending
> the single-head paradigm in a logical and consistent way.

|Well, I think we have very different working environments. I

I think that is the single most important thing for anyone to
keep in mind when working on end-user computer software :-)

|prefer to have 5 clients at maximum in my view, otherwise I
|cannot concentrate on what's going on anymore. Sometimes I throw
|a glance on what's going on in a different view through toggling
|a tag, but that's the only scenario when I have more clients
|than 5 around.

Firstly, I'll mention something about my mind that might be important to
understanding how I work: when I'm trying to figure out
some discrepancy/relation between two windows, eg, source
files, I _really_ have problems concentrating and picking up
differences/relations by changing quickly between the two in one
window (eg, swapping between buffers in an emacs window). I find it much easier to
have the two windows up on the screen and alter between
looking at the two. (Eg, having a C++ header file
containing an interface definition and a C++ implementation
file up, or a program and a textual log of the output it created,
or a program and a graph plot it created.)

Now, I work in academic research on computer vision and I
find that _in my particular work environment_ I find I spend most of
my time trying to fix things caused by inconsistent defintions between
different places, or trying to relate some feature of a graph plot
with the code of an algorithm that generates it, or some other
activity involving looking at two or more windows at once. I'll
often have between four and ten windows in a view and they'll
all be showing "aspects" of the same task, even though they might
be in different applications (eg, editor windows, terminal windows,
graph plotter, image/movie display, etc). There's a similar sort of
thing going on when I write papers, where there's a main text
window, bibliography window, graph window, pdf viewer window.
Again, one task but multiple applications.

I should note that it's not that I've got wildly dissimilar applications like
editor, a web-browser, an irc client, a TV application and a game
all in one view: I'd agree no-one can concentrate like that.
At the back of my mind is always how I work when I'm doing
some paperwork on my physical desk: I quite often have a several
papers spread out and move my gaze between them rather than
move whatever I'm currently working on to the physical centre of the desk.
Likewise, having multiple clients really just means that I can move
between them just by moving my gaze rather than having to engage
my brain to figure out how to change what's viewed in either
the application or window manager.

(Note that I also have a laptop I frequently use on a train, where I use
a more standard dwm usage style.)

Anyway, that's _my_ view on things but the meta-point was that
if you're thinking about multi-head as someone who doesn't use
it, I do think it'd be useful to get
feedback from others who do about any human-issues that arise,
because they aren't obvious.

|In a setup with many windows opened, I imagine you need a
|totally different approach, but I doubt this scales well with the
|design decisions in dwm, out of curiosity, did you tried

As mentioned above, I've got a mix of source code windows
where "tall and relatively narrow" is desired and image/graph
windows where you want "wider than tall" so being able to
put them into a "slave" column makes sense. So I didn't try a
"make them all the same shape"-type grid layout. The most
awkward thing Icoulde see is that X doesn't (AFAIK) provide any
way for an application to specify that it has a desired
_aspect ratio_, so I started to imagine all sorts of hacks in a WM
for figuring out which clients wanted exactly the aspect
ratio implied by their initial geometry and which were truly
resizable. And then I saw the chasm of hard-to-debug complexity opening
up in front of me so I abandoned that idea.

cheers, dave tweed
Received on Thu Sep 28 2006 - 19:44:25 UTC

This archive was generated by hypermail 2.2.0 : Sun Jul 13 2008 - 14:31:39 UTC