From: Guido S. <gu...@ar...> - 2004-01-31 13:18:32
|
Am 30.01.2004 19:25:17 schrieb(en) Stephen Watson: > Guido Schimmels <gu...@ar...> wrote: >=20 > > Am 26.01.2004 18:32:47 schrieb(en) Jonatan Liljedahl: >=20 > > > /Jonatan - who would love a standalone ROX-WM! > > > > If you want a WM for ROX, there is no reason why it shouldn't be > part > > of the filer, >=20 > Let me quote the front page of the ROX web site: "The ROX style > favours > using several small programs together instead of creating all-in-one > mega-applications." I had the idea originally when playing with the WindowLab WM, which =20 compiles into merely 30k, a taskbar included. Unfortunately nowadays =20 with increased requirements and the growing pile of WM hints, it seems =20 you don't get away with so little WM. Apple's OroborusX fork let me have a look at the Oroborus WM, which =20 compiles to 80k. Considerably more, but still not that bad. And I =20 already see how some 30% of it's source files can be dumped due to =20 overlaps with the ROX-Filer sources. You can not compare a jump from =20 450k to 500k-550k to monstrosities like Evolution, OpenOffice or =20 Mozilla. If we were talking 200-300k for the WM, then I'd agree, =20 obviously a bad idea. And a window manager is not really an application in the sense of a =20 addressbook, calendar or mail client anyway. > ROX-Filer is a file manager, not a window manager. ROX-Filer with its pinboard and panel is more than a file-manager =20 already. With an integrated WM it would be a complete desktop shell. Window Managers are a X11 curiosity. X was originally designed as an =20 academic toy for experimentation with graphical interfaces. It's not =20 the way it is because it made actual sense. I might be wrong, but I =20 don't think you'll find something like a window manager process =20 anywhere else. The task of a window manager is trivial, the fuss made =20 about it by Linux users pathetic. Really, OS/2 Worplace Shell, explorer.exe, BeOS Tracker. There is =20 nothing odball about this kind of integration. It doesn't necessary =20 mean the "desktop" is best implemented as a single binary. But it =20 definitely should be presented as such to the user. > > but many reasons in favour: > > > > - reduced memory usage > > - fewer context switches > > - reduced start-up time from the login prompt >=20 > Trivial gains I would say. > > - usability: ROX-Filer->Options->Window Settings > > (if you think this is not an obvious place to look, open ROX-Filer- > > Options now, dig through the tabs and count the word "window") >=20 > The relevant options being: > * Handle iconified windows for the window manager, a side effect of > the > pinboard > * Compatability options, which will go away eventually when the WM's > get > fixed. >=20 > > - 100% compatibility with ROX-Filer guaranteed: better chance =20 > users >=20 > > arrive at a working setup; especially important as ROX is a > neglected > > step-child of all major distros >=20 > I've never had compatability problems with Sawfish. Obviously other > people > have had trouble with their WM, From the 100 or so WM's maybe 5 are fully compatible with ROX-Filer. The problem is all of those WM's are in use and people try to use them =20 with ROX-Filer. And I'm so tired and bored about the matter that I want =20 to make sure I'll never have to answer such questions for RoxOS. It's =20 so old. > but only with panels and pinboards > which you > were deriding earlier. That's a miss-understanding. I wasn't deriding them. Just pointing out =20 the irrational resistence the possible wm integration met. There may be =20 sound technical reason against it like reduced responsiveness . But =20 that's a different story. > I really think it's a lot of code to add and maintain for so little > gain. No foreign WM will ever fully meet the needs of RoxOS. So this is =20 really only a matter of code organisation. Things like pinboard item =20 accelerators can not fully work when the filer doesn't know which ones =20 are already taken. And for "a lot of code" see above. How much WM does =20 one really need? |