Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!wuarchive!decwrl!shelby!helens!baroque!jim From: jim@baroque.Stanford.EDU (James Helman) Newsgroups: comp.sys.sgi Subject: Re: SGI's X (Re: X on the Personal Iris) Message-ID: Date: 14 Feb 90 19:30:33 GMT References: <22195@uhnix1.uh.edu> <22066@pasteur.Berkeley.EDU> Sender: news@helens.STANFORD.EDU Distribution: usa Organization: Stanford University Lines: 56 In-reply-to: seth@miro.Berkeley.EDU's message of 13 Feb 90 19:29:48 GMT Ahh luv the smell of napalm in the morning. Be forewarned, I even start my barbie with a blow torch. Seriously. I find it is much more effective than lighter fluid. I won't go into the litany of the "non-specific" problems we've had in several areas. But they were substantial and have a long history. Now that I know where I can send them, I've submitted them to SGI. Hopefully, resolution is just a short release away. I would like to share one divine revelation though. The New Testament says there is one sin which can never be forgiven, but never names it. But, it has been revealed to me. The blackest depths of hell are reserved for those who violate The First Law of Window Systems: A window server shall never dump core except in case of hardware failure or nuclear war. I suppose, DOD might take exception with the latter exclusion. But anyone who has used buggy window systems will agree, that when a window server dies, it's a very bad thing. All that context, jobs in progress, network connections, my very stateful 25MB of LISP and data with two hours of processing time, all go down the toilet. If it happens enough times, it can adversely affect your personality. "Experimental" stuff like MIT's X10R3 server dumped core rather frequently. As did MIT's X11R2 before all the patches made it out. We completely skipped X11R1 because of its reputation. But, at least on our Sun-3's, I don't think I ever got dumped on by MIT's X10R4, X11R3 or X11R4 server. Not bad for code that isn't even commercially supported. Similarly SunView (ok. so it's not a server.) never bombed on me. The same can not be said of any of SGI's X releases to date, although, as I said they are getting better: the MTBD is down by at least factor of ten. Similarly, in all prior 4Sight releases (I haven't seen it yet in 3.2), if you tickled GL the wrong way or even did normal operations on a heavily paging system, grcond would choke, and you would get bombed back to the login prompt. Miss Manners would not approve. As I said about SGI's X. It is now faster and more usable than ever before. The OS is incomparably better than in the IRIS 1400 days, aka the stone age. But if Silicon Graphics aspires to be a big guy, (in the long term, there will only be big guys in the workstation market), it needs not only to offer leading edge performance (which they do better than anyone else) but also offer absolutely top notch software and support. Jim Helman Department of Applied Physics P.O. Box 10494 Stanford University Stanford, CA 94309 (jim@thrush.stanford.edu) (415) 723-4940 Waiting for perfection, in the meantime: "Man is not at all good. So hit him on the head. If you hit him on the head enough, then perhaps he will become good" - Brecht, The Threepenny Opera