Re: gnome-terminal idea

[Tim Moore <tmoore tembel org>]

On Wed, 23 Sep 1998, George wrote:

> On Wed, Sep 23, 1998 at 11:24:30AM -0400, Tim Moore wrote:
> > > I don't know why anybody would want to use anythign else then vi ...
> > > does that mean that all editors should be erradicated and vi should be used
> > > for all
> > 
> > Well, I can think of lots of reasong why someone wouldn't want to use
> > vi, but We won't get into that. :-)
> 
> that's not the point ...

I know. The point is that people should be accepting of other ways to do
things. However I want to go beyond accepting and try to *understand* why
people want to do things in other ways. Partly because I might learn of
better ways to do what I want to do. Partly because I might be able to
find a need that wants to be served.

For example, consider that there are a lot of people out there that use
Microsoft Windows. Obviously, then, a lot of people prefer it to Linux. I
personally think Linux is superior, but it is obviously not superior *in
every way.* On the one hand I could say that people are different and if
they like Windows then I shouldn't concern myself with it and its obvious
shortcomings. On the other hand I *do* say that everyone should be using
Linux and so I need to work to eliminate the reasons they have for
preferring Windows. The eventual goal, IMO, *is* to eliminate Windows --
not by restriction (i.e., not letting people have what they want) but by
incorporating the advantages which it has as much as possible.

So back to the situation at hand: I saw a lot of problems with the
GNOME-MDI interface and didn't really see the appeal. That's why I posted
the original message; not to offend those who like it, but to understand
its advantages. And that's why I'm arguing against it, to get a more
precise understanding of what makes it a superior solution for some
people. Then perhaps a solution can be found which has those advantages
and avoids those drawbacks.

> 
> > But I do think that if nobody actually uses some feature or app, it
> > probably *should* be eliminated because its a waste of effort and
> > clutters the interface. I'm not trying to suggest that this is one of
> > those cases, but I do think there are better ways to accomplish the same
> > thing.
> 
> just because it got implemented can already give you a hint that someone
> thought it was usefull and is most likely using it ... (most free software
> people write, they also use) ...
> 
> it's already there and there is definately at least one person who likes it,
> (e.g. me) ... 

I realize all this. Like I said, this is not one of those cases. It may
become one, however, if we can determine a better way to confer the
advantages of the tab approach.

> elimination of stuff that you think is unuseful is actually pretty dumb ...
> unless you are sure that noone is using it 

Well, perhaps elimination was the wrong term, but the idea is not to put
effort into things that you think are useless (which is pretty obvious) 
and also to put effort *against* things you think are counter-useful. Now
GNOME-MDI isn't *that* counter-useful which is why I'm not putting *that*
much effort against it. I'm more trying to evaluate what makes it useful
and to whom and whether the problems it solves can be addressed by
something that doesn't have the problems it creates.

> in this case .. it's NOT the case ... I for one am using a wm without a
> taskbar .. and I don't really like taskbars ... so I would like the
> mdi as a notebook ... I bet tehre are a lot of people that don't use
> taskbars ... or that find the tabs better then a taskbar ....

So, hypothetically, if your window manager supported the BeOS-inspired
docking tabs interface, would you prefer that or the GNOME notebook and
why?

> 
> > > taskbar includes all the windows, tehrefor it's quite confusing to find
> > > the window you want unless you keep your windowcount to a minimum,
> > > I often have several netscape windows, and on each desktop at least a
> > > couple of open terminals ... if I could have a window with tabs on each
> > > desktop I would save so much space, while still making it easy to switch
> > > between different windows ...
> > 
> > Some WMs I know only put the windows on the current desktop in the
> > taskbar, which I think is the effect you're describing.
> 
> what if I have 10 rxvt's open and two vim windows both with about 3 files
> in it, all on one desktop ... this would mean 12 windows in the taskbar ...
> or 16 if vim used your form of mdi (elvis does for example) ... hmmm ...
> that's not good ... I want a quick way of switching between the 3 files

I think I see your point: Having all of the documents in a single "switch
space" means that you would have to cycle through a large number of
windows to get to the one you want, and the taskbar would fill up. So you
want kind of a two-level nesting so that one switching method would switch
between apps, and the other would switch between documents within an app.
I can see the appeal of this, but I don't think it's the best way to solve
the problem:

Are 10 tabs really more manageable than 12 taskbar buttons? And what if I
have 20 browser windows open and no other apps on the desktop (which I
sometimes do)? The problem has shifted from a cluttered taskbar to a
cluttered row of tabs! 

What if I'm working with a picture in GIMP, two HTML files in some text
editor, and previewing the page in a web browser. This two-level grouping
doesn't correspond to my task in any way, and in fact may get in my way
when I'm trying to rapidly switch between these four windows, and have to
go through two different routes. It's even more pronounced if I'm trying
to work on two tasks at once. Say, in addition to the web page task, I'm
working on some software. To do that I have another two text files open, a
debugger window, and a terminal window where I'm compiling stuff. While
I'm waiting for the compile, I work on the web page. In this situation, I
want to be able to switch between the two tasks occasionally, and the
windows within each task rapidly and often. This roughly corresponds to
the app/document nesting (Alan Cooper calls this kind of two-level nesting
a monocline grouping) but I don't care about applications -- the HTML
source and the C source aren't really related in my mind. I care about
what I'm trying to do.

If the tabs worked like this, I may be more inclined to use them.
Basically you would have to have a generic container window which could
contain tabs from multiple apps. I could dock arbitrary windows together
and split them apart whenever I feel like it. In fact, I think someone
made a proposal to this effect, calling it an Emacs-like MDI (it's on the
GNOME proposals web page IIRC).

However, this has its own problems. Any crashing app could bring down the
entire notebook. And the notebook-dock itself would have to be perfectly
stable. And it would have to have all of the notebook-tab management
commands I would ever need. Essentially, it would take over a lot of the
task of window management, turning it into notebook-tab management.

So if the ideal is to have a generalized window-docking feature, I would
consider this the domain of the window manager (hopefully Raster is
reading this!) IMO, implementing the BeOS-inspired docking windows will
solve all of the same problems that GNOME-MDI does, plus some.

On the other hand, I could be missing something. There may be some reason
why grouping by application (and by application process, at that, rather
than all running instances of the application being combined) is better
than arbitrary user-defined grouping. There may be some reason why this
should be handled by each application instead of by the window manager. If
so, enlighten me!

> > OK. I can agree with that of course. I still think there are better ways
> > to deal with the problem than combining all windows into one. David
> > Jeske's idea to have "dockable" windows with tabs for titlebars is one.
> 
> I kind of like the current implementation ... maybe there could be
> improvements to parts of it ... but otherwise it sounds fairly reasonable to
> me

Is there anything you don't like about David Jeske's idea? What benefits
does the GNOME idiom have over the BeOS-inspired one?

Tim