How I use computers
I have posted a review of Gnome 3, and will shortly post a review of Gnome 3.2, where I indicate my dislike for the new versions of gnome. So I thought it might be useful if I described how I typically use computers, and how that relates to my choice of desktop environment.
I have used slackware with fvwm (various versions), redhat 7.3 with KDE, solaris 7, solaris 8 and solaris 10 with CDE, SuSE 9.x (I think it was 9.1, but my memory is hazy) with KDE, SuSE 10.1 and openSUSE 11.0 with Gnome, openSUSE 11.3, 11.4 and 12.1 with KDE4. And I have also experimented with XFCE, LXDE, Gnome 2, Gnome 3, icewm on those later openSUSE systems. The way that I have used the systems has been much the same throughout, except with Gnome 3, which gets in the way of how I want to use the system.
I’ll describe my current setup, which I am using with openSUSE 11.4 and KDE 4.6.0. My setup with other systems was similar.
I use 4 workspaces or virtual desktops. I don’t give them names – the default numbering works fine.
I start with an xterm session, which I make visible on all desktops. I use that when I want to do something from the command line.
I use Desktop 1, mainly for term use (I prefer xterm to other terminal emulators), and this often includes ssh into remote systems. I use Desktop 2 for web browsing, and related activities. I use Desktop 3 for a GUI email client (currently claws-mail, though I have used Thunderbird and exmh in the past). I actually do most email with the browser via webmail access, and I do some with command line email clients (often “nmh”). I use Desktop 4 to edit latex files, compile them, browse the resulting dvi file (with xdvi).
Having these kinds of activities on different desktops makes it easy to switch between them when multitasking. I am not always doing all four. Often, I am only using Desktop 1 and Desktop 2. But I use the others often enough that it is worth the effort of putting them on separate desktops.
I configure the system for “focus follows mouse” or “sloppy focus” (different desktops describe this and implement it a bit differently). I find that more flexible and more powerful than “click to focus”. I also configure the system so that panels are set to auto-hide. That makes available the maximum screen real estate for use in the various applications.
What I have described works out fairly well in all of the desktops that I have mentioned, except Gnome 3.
With LXDE and icewm, I did not find a way of configuring the system to start with that single starting xterm. So I do have to manually start that, and set it to be visible on all desktops. Back when using fvwm, I did that a bit differently, putting startup windows in the “.xsession” file that I used. With those exceptions I can set things up the way I want them, except in Gnome 3.
With Gnome 3, normal mode, the desktops (or workspaces) are dynamic. They are created as needed, and disappear as soon as the last application is closed. That makes it impossible to have a standard place to look for where I am doing latex or reading email. Additionally, I have not found a way to auto-hide the panel with Gnome 3, and that reduces the amount of screen space available. I mostly notice that shortage of screen space with the browser and with xdvi, where it is useful to have as much vertical space as possible. Setting “focus follows mouse” also turned out to be hard, though some web searching showed how.
If I switch to fallback mode in Gnome 3, then things are better. I can still have my 4 desktops, and have them stay where they are. But the downside is that there is now a bottom panel as well as a top panel, for more loss of screen space.
If I were a tablet user, then I might take a different view. The tablet computers don’t seem well suited to the use of latex nor to ssh logins to remote hosts. So I would probably be using the tablet mainly for web browsing, for use as an ebook reader, and for multimedia. Gnome 3 probably works quite well for that range of applications.