mo@maximo.UUCP (Mike O'Dell) (02/05/88)
The astonishing baroqueness of X is the greatest threat to the general sucess of UNIX to have come along since System V hit the streets. If you try to give an X system to a real human being, not a computer hacker masquerading as a normal person, they will croak. If X doesn't instantly burn out their eyes and brain, causing them to throw their UNIX box out the nearest high window, it will drive them straight into the arms of the Macintosh II. With the toolbox under AUX, all the windowy programs on the MacII will have a clear, understable, and universal user interface. With other alternatives, we face the very real prospect of each window (program) having a different user interface. That, friends, will be the death of UNIX. The Ol' Curmudgeon -Mike O'Dell "Nature neither seeks nor abides opinions."
bzs@bu-cs.BU.EDU (Barry Shein) (02/06/88)
Posting-Front-End: GNU Emacs 18.41.4 of Mon Mar 23 1987 on bu-cs (berkeley-unix) >The astonishing baroqueness of X is the greatest threat >to the general sucess of UNIX to have come along since >System V hit the streets. If you try to give an X >system to a real human being, not a computer hacker >masquerading as a normal person, they will croak. > The Ol' Curmudgeon > -Mike O'Dell Although Mike is not that old I tend to otherwise agree with him. The problem I see right now is that the "truisms" run something like this: 1. X is a protocol definition, the code people are looking at right now is a particular implementation of that protocol, the two thoughts should not be confused. 2. The ToolKits will fix the programmer's interface complexity. 3. No, well, actually, the Toolkits will not fix that problem, some sort of as yet unspecified 12th generation point and click programming interface will fix the problem (this was after it seemed to be generally agreed that less than 5 people in America can understand the Toolkit supplied, and they're arguing amongst themselves about whether or not it can ever work, and besides, it only handles a small subset of X anyhow and is too complicated. And worse, although it's implemented in C for C programmers it was never intended to be, they were forced to by someone, it should have been implemented in some as yet unspecified language which would solve *everything* according to the designers.) Somewhere in here I sense a wheel of reincarnation. If it's only a protocol and we should ignore (well, only if we don't like it) the current implementations of those protocols then I suppose we better not program anything as our code is doomed to obsolescence right after the next major semester break at MIT. If we shouldn't be coding at the level of XLIB and be using Toolkits instead as they provide the abstract interface we all desire then how come people privy seem to agree that the supplied toolkit is basically incomprehensible and should be, we should be waiting for something else? It's all starting to sound like Nixon's Secret Plan to end the Vietnam War (I suppose that metaphor reveals me to be as old as Mike.) The wheel of reincarnation reference is that if someone comes along with a toolkit which is useful then the first thing we should do is declare that to be X and throw away anything "below" it other than the protocol. If it can't do that then it's a pretty poor toolkit (not powerful enough or something.) Of course, then we will ask, why wasn't that approach taken in the first place? Well, I suppose one can say, because hindsight is 20/20. The road to hell is paved with myopia. Disclaimer: I sort of like X11, I use it as my primary window manager on my Sun, have ported the client interface to an Encore Multimax. Clients I have written are "out there" (mostly X10.) I recently wrote an X11 plot interface to DOE MacSyma which should appear on their next release tape and am involved with various groups' alpha and beta-testing new clients and I am working on some of my own. What I like about X is that it is available in a form which basically works (I'm not sure I can say the same for the competition yet) and is available under a very agreeable source release. Unfortunately there may be fundamental flaws in the model (eg. moving to different res or other variation devices seems to be very painful, again, that may not be the protocol but the implementation, as the NewSpeak goes.) I suppose one might say that right now there is nothing approaching a standard window system for Unix. NeWS might be a competitor some day although it may have fundamental flaws also (not so much in the windowing model which is very good but in the implementation approach.) Basically, X and NeWS seem to form the right and left brain halves of windowing systems. X is basic, fast (or should be) and analytical, NeWS seems to be what you should be using if you want something more creative than boxes with chars and/or line drawings in them of a fairly fixed nature. Right now people seem to be responding to each on that atavistic level. X is a jeep wagoneer with all options including a tow ball if you can't fit it inside the cab, NeWS is a DeLorean turning magnificently on a stand in the main lobby of the Museum of Modern Art, the engine comes in kit form, diesel, gasoline, ethanol, any number of cylinders all available, actually the kit is just a big cube of steel, very high grade, and a textbook on modern engine design. The X11/NeWS merge might very well end up to be the "long-awaited" station wagon version of the DeLorean, with the jeep hanging off the back on a newly attached brushed stainless steel tow ball, just in case. -Barry Shein, Boston University
gilbert@hci.hw.ac.uk (Gilbert Cockton) (02/12/88)
In article <8802032048.AA12237@uunet.UU.NET> mo@maximo.UUCP (Mike O'Dell) writes: >The astonishing baroqueness of X is the greatest threat >to the general sucess of UNIX to have come along since ... There is still hope for a decent X tool-kit. Why can't the MITechies look at the development of Star and the Macintosh? Two sets of skills played a major role in their attractiveness: human factors evaluation and graphic design. The latter may be more responsible for the attractiveness of the Mac than anything else - after all once you start using a Mac, you realise that the HF evaluation wasn't always too searching. Yet the Mac attracts users who run miles from real computers :-) - SunView, GEM, AMIGA toolkit and MSWindows don't seem to have been anywhere near a graphic designer either. So why don't MIT put a little of that SciFi research money into designing a toolbox *PROPERLY*? -- Gilbert Cockton, Scottish HCI Centre, Heriot-Watt University, Chambers St., Edinburgh, EH1 1HX. JANET: gilbert@uk.ac.hw.hci ARPA: gilbert%hci.hw.ac.uk@cs.ucl.ac.uk UUCP: ..{backbone}!mcvax!ukc!hci!gilbert