gancarz@decvax.UUCP (Mike Gancarz) (06/20/87)
I'd thought we'd gone through all this once before. Anyways... In article <> Karl.Kleinpaste@cbstr1.att.com writes: >No, thanx; give me the naked mouse hits any day. This will lose big in certain situations, such as when windows are larger than the screen, the root window is occluded, the screen context obscures your "grow boxes", and so on. Of course, since X allows you to run multiple window managers concurrently, you can always use xnwm when you've got a Dr. Pepper in your hand and uwm when you have both hands free. :-) --Mike
gancarz@DECVAX.DEC.COM (Mike Gancarz) (06/22/87)
Charles Haynes stated that: >It *is* possible, with a little bit more work, to make your window >manager context sensitive. This allows "naked" mouse hits to be >meaningful in the root, and in icons, while not superseding them for >applications. This provides the best of both worlds. I agree. The window manager should listen to the root window instead of doing grabs unless root, window, and icon contexts have all been specified for a given button mask. This is a bug in uwm and it should be fixed. Please forward your suggestion to Loretta (Guarino-Reid) as she currently maintains uwm. --Mike
haynes@DECWRL.DEC.COM (06/23/87)
I'd thought we'd gone through all this once before. Anyways... In article <> Karl.Kleinpaste@cbstr1.att.com writes: >No, thanx; give me the naked mouse hits any day. This will lose big in certain situations, such as when windows are larger than the screen, the root window is occluded, the screen context obscures your "grow boxes", and so on. Of course, since X allows you to run multiple window managers concurrently, you can always use xnwm when you've got a Dr. Pepper in your hand and uwm when you have both hands free. :-) --Mike I don't know about "we", I know *I've* been through this, and I am continually annoyed by the fact that if I allocate "naked" mouse clicks for de-iconify, I lose them for ALL applications, ALL the time. THIS IS ANTI-SOCIAL BEHAVIOR. It *is* possible, with a little bit more work, to make your window manager context sensitive. This allows "naked" mouse hits to be meaningful in the root, and in icons, while not superseding them for applications. This provides the best of both worlds. Furthermore, in X11 it is possible to make "naked" mouse hits meaningful to both the application AND the window manager in the same window. Of course it means the window manager writer needs to be more careful in the design and implementation of the window manager, but it is doable. MOUSE BUTTON GRABS ARE EVIL. They should be used with utmost care and only when ABSOLUTELY necessary. -- Charles
Karl.Kleinpaste@cbstr1.att.com (06/23/87)
Posting-Front-End: GNU Emacs 18.47.3 of Sat Jun 20 1987 on cbstr1 (usg-unix-v) First, I am sorry if I caused a rehash of a rehash of a rehash of an issue. That was not my intent, and I hope this doesn't continue it. Second, the one reason for my posting was in the one quoted line of its precedent posting, which said > The other part of the reason [for the existing defaults] > was to avoid overwhelming the novice user. My followup article questioned only this one claim: that the current set of defaults (requiring 2-handed X driving) is optimal for overwhelm-avoidance in the novice's mind. I am not arguing against the defaults *in general* because I understand that [a] uwm just happens to be the existing window manager and one could write another and [b] uwm is highly configurable and hence the defaults are not engraved in stone. My lone, sole, single, isolated complaint is that 2-handed driving for a window system, particularly on machines where other window systems exist (such as Suns, of course), is a bad set of defaults to inflict on the *novice*. OS experts are expected to get used to the intricacies of kernels and the more esoteric utilities available (such as, say, awk in UNIX). X experts can similarly be expected to manage with bizarre and not necessarily intuitive arrangements of mouse/keyboard hits. But you don't hand a UNIX beginner anything tougher than ed(1) for an editor, and you shouldn't hand an X beginner a mouse that has to be metafied. And to think that my followup was written 3 weeks ago, but got stuck in someone's buggy news system in Columbus...sheesh. Karl
jdm@gssc.UUCP (John D. Miller) (06/25/87)
personally, i kind of like the approach that the "layers" window manager
uses, with "arbitrary" pop-up menus that are activated by button hits, rather
than position. click button #N anywhere and a system menu comes up at that
location; button #N+1 may behave differently, depending on where the mouse
currently resides (i.e. an application window).
this scheme would be even better if the global button grabs could be
selectively released on some or all clients, such that while running the
foobar paint program, i could release or remap button #N (mentioned above) to
the paint program that needed it, but elsewhere would behave the same. of
course, the foobar paint program would only grab button #N when it needed it...
--
-- jdm "Caution, sleep turbulence."
in real life: John D. Miller, Graphic Software Systems (GSS), Beaverton, OR
...!{tektronix!verdix}!sequent!gssc!jdm (503) 641-2200gancarz@decvax.UUCP (Mike Gancarz) (06/25/87)
In article <> Karl.Kleinpaste@cbstr1.att.com writes: >My followup article questioned only this one claim: that the current >set of defaults (requiring 2-handed X driving) is optimal for >overwhelm-avoidance in the novice's mind. I would tend to agree with you that the current set of defaults in uwm is probably less than optimal. In preliminary testing, however, it became evident fairly quickly that "naked" mouse bindings in uwm tended to confuse novices because they have a hard time dealing with the fact that they couldn't use naked bindings in any applications. Once you tell them "hold down this key when you want to call up the menu", they usually catch on rapidly. To do the right thing here (whatever *that* means), you'd have to perform a human factors study with at least a dozen default environments and hundreds of users. And whatever the study came up with, I'd give you 100-1 odds that the standard default environment won't please everyone. Hence, the desirability of a user-programmable user interface. >My lone, sole, single, isolated complaint is that 2-handed driving for >a window system, particularly on machines where other window systems >exist (such as Suns, of course), is a bad set of defaults to inflict >on the *novice*. What you're hinting at is that, since someone else has already done it one way, then that way must be right. Or is it? Should all window system user interfaces look alike? That's a tough question. Someone with lots of experience in the window system arena once put it this way: "Don't say your way is better because you've got an army of people who say your way is better. I can go out and find an army of people who would disagree with you." If X had come along before the Suns, the Xeroxes, and the Macintoshes of this world, then maybe everyone would be wondering why those systems don't use meta keys for window management. (By the way, did you know that the Macintosh uses meta keys in its user interface? Surprised? I was, too.) --Mike
Karl.Kleinpaste@cbstr1.att.com (06/25/87)
Posting-Front-End: GNU Emacs 18.47.3 of Sat Jun 20 1987 on cbstr1 (usg-unix-v) gancarz@decvax.UUCP writes: > I'd > give you 100-1 odds that the standard default environment won't > please everyone. Hence, the desirability of a user-programmable > user interface. Certainly. No one expects to please the world. But one would do well to introduce the world to something new in a more gentle fashion. Perhaps 2 sets of defaults would have been appropriate, one for naive novices, and one for experienced users who need a base from which to customize. Just a thought, and it might be a bad thought at that. > What you're hinting at is that, since someone else has already done > it one way, then that way must be right. Or is it? Not at all. The only thing I'm arguing for is the Principle of Least Astonishment. The conditioned user of another window system walks up to an X display, tries to do a few things, can't seem to get much out of it, and then a local expert tells him to metafy a mouse button. The novice says, "I have to WHAT?" P-of-LA violation. He's *used* to naked mouse bindings, so give him what he's *used* to while at the same time showing him what he could do if he'd get used to metafying those buttons. It seems you're giving the novice the full power of uwm all at once. Can the power be stepped down sanely? If X had been the first window system, then users would have become conditioned to it first, and the question they'd ask when walking up to the next window system is, "OK, which meta/shift/ctrl keys do what with this mouse?" X has the disadvantage that it was not first. That is also an advantage in that it has learned from many of the mistakes of the past. > (By the way, did you know that the Macintosh uses meta keys in its user > interface? Surprised? I was, too.) Yes, I found out about that quite a while back. And as one who seldom uses a Mac, I don't care for it - but again that's just me, personally, a Mac novice. Karl
gancarz@decvax.UUCP (Mike Gancarz) (06/26/87)
In article <> Karl.Kleinpaste@cbstr1.att.com writes: > >Certainly. No one expects to please the world. But one would do well >to introduce the world to something new in a more gentle fashion. We are in violent agreement on this point. Like U*IX, uwm was not intended for use by novices. But you know how it goes... >Perhaps 2 sets of defaults would have been appropriate, one for naive >novices, and one for experienced users who need a base from which to >customize. Just a thought, and it might be a bad thought at that. 2 or more sets of defaults would be more appropriate. Novices and experts use systems in vastly different ways, so their user interfaces should be varied accordingly. I would nominate xnwm in "menu bar" mode as a usable novice interface and uwm as a reasonable interface for the seasoned user. Both would need lots of work to fulfill their respective ends of the spectrum, however. >Not at all. The only thing I'm arguing for is the Principle of Least >Astonishment. X itself breaks this principle in so many areas, I don't see why its user interface(s) should be an exception. :-) --Mike
hugo@gnome.cs.cmu.edu (Peter Su) (07/01/87)
> >> (By the way, did you know that the Macintosh uses meta keys in its user >> interface? Surprised? I was, too.) > >Yes, I found out about that quite a while back. And as one who seldom >uses a Mac, I don't care for it - but again that's just me, >personally, a Mac novice. > >Karl One should point out though, that the Mac uses meta keys only for very special operations, or quick hacks. You don't need to use them to do "normal" things like moving windows about and stuff. Generally, Mac programs use the shift/option/command keys for shortcuts, like option-click in the close box closes all open windows when in the finder sorts of things. Pete -- ARPA: hugo@gnome.cs.cmu.edu UUCP: ...!{ucbvax,ihnp4,cmucspt}!hugo@gnome.cs.cmu.edu "There are reports that many executives make their decisions by flipping coins or by throwing darts, etc. It is also rumored that some college professors prepare their grades on such a basis." - Donald Knuth