A Desktop Environment, like a shell, is a user interface to an operating system, but whereas a shell provides a linear interface, a desktop environment provides a two or three-dimensional interface.
On comp.lang.tcl, Victor Wagner wrote:
Now, I'm suggesting a technical challenge for people who want to get Tcl into headlines of computer-related news:
Write a desktop environment in Tcl.
If it is properly done, we'll get something uncomparable with Windows, KDE, and MacOS - an easily extensible desktop environment, where user has powerful scripting language at his disposal. With Tcl introspecting capabilities it is even possible to write RAD tools which would allow to do nontrivial scripting by drag and drop, and resort to typing commands in to enhance drag-n-dropped prototype.
So we should write fully scriptable suite of components, which would allow GUI user to extend and tune his environment even simpler than a command-line Unix user can tune his environment using shell/perl/tcl scripts, aliases and such.
We already has almost all necessary components - web browser (BrowseX), spreadsheet (Abacus), window manager (tkwm) etc, etc.
We also have ways to integrate traditional Unix tools into our graphic apps (expect)
All which is needed is to write proper integration framework.
It is sufficient to run all the components of desktop in single nonthreaded process. MacOS did this for decades, and most users were satisfied. No CORBA would provide anything similar by flexibility to interface, Tcl extension mechanism is providing. But of course, CORBA should be supported to communicate with foreign apps.
But we also have powerful means of interprocess communication (Tk send) which allows communication even between programs running on different machines, as long as they are on same display (which is proper desktop approach, I think).
So project may be finished with fewer man-months than it is already invested into KDE or GNOME.
What we need:
I would like to see a desktop development framework provide:
Anselm Lingnau said:
It's true that there are lots of pieces around. However a »desktop environment« to rival the likes of KDE would probably require more consistency between the individual pieces, not only as far as the »look and feel« is concerned, but also for smooth interoperability.
It would be nice to have an »application framework« that would make it easy to generate the skeleton of a »desktop-ready« Tk program. Right now there is quite a lot of drudgery required to set a program up with a menu bar, »help« menu, toolbar et cetera (»drudgery« from the point of view of a Tcl programmer -- it is still a lot more straightforward than with other toolkits). It would be nice to be able to say something like
package require tkapp
set menu {
File {
command &New [list newCmd]
command {&Open ...} [list openCmd]
command &Save [list saveCmd]
command {S&ave As ...} [list saveCmd -ask]
-
command &Quit [list quitCmd]
}
Edit {
...
}
}
set toolbar [list
[tkapp::tbicon new] newCmd
[tkapp::tbicon open] openCmd
...
]
set about {
Foobar 1.0
by J. Random Hacker
See http://www.foobar.org/
}
tkapp::create -name foobar -title Foobar -icon foobar.xbm \
-menu $menu -toolbar $toolbar -about $about \
-manual [file join [file dirname [info script]] doc]where the »tkapp::create« procedure does all the work of installing the top-level menu, toolbar, »About« dialog and link to documentation viewer, reads the resource file(s), sets the main window title and so on. This would go a long way towards enabling a common look-and-feel for Tk »desktop« programs, since suitable definitions could be pulled in without making extra work for the application programmer. It would even be possible to make use of, say, existing KDE preferences files to find out about global settings for fonts etc., the way Ktk does this. In any case, the »tkapp::create« procedure could also try to do the right thing on other platforms such as KDE, GNOME or Windows.
Does a brand new desktop environment need built - one that works cross platform? Or is it more important to work on integrating Tk/Tcl better into the existing desktops, given that right now people complain about the behavior and look of tcl/tk on gnome, kde, MacOS X, Windows XP, etc.?
There is no way that we can compete in this space. The number of committers to Gnome and KDE is probably larger by an order of magnitude than to Tcl, Tk and the projects that comprise the "activestate distribution". The most sensible thing is, as the above comment states, to try and integrate with existing environments in order to take advantage of the huge amount of work they have done.
SRIV 2004-08-28: I disagree. The RAD nature of Tcl empowers the developer to create powerful apps in 1/10 the time. Other than having to play catch-up with other desktop environments, a tcl desktop would have a significant advantage. By utilizing the "app framework" methodology, integrated apps could be easily created or converted from old apps. This project has a lot of potential and should not be discounted so easily. I can imagine a tcl desktop stored on a USB keydrive that runs on top of any windows/linux window manager, which gives a unified set of apps and functionality. Try that with KDE or Gnome!
Rildo Pragana 2005-03-02: I think it is very viable to have a Tcl desktop (with the help of a small extension). I'm still doing some experiments, but you can already see what my experiments have led me to, at http://pragana.net/ and grab a starkit of it for Linux. It will feature a file manager, taskbar, desktop, fancy root menu, and (not yet started) a window manager. The source of the small C extension is also inside the starkit, so use "sdx" to access it, or the included file manager (browse into the virtual directory with the name of the executable).