Return to BSD News archive
Xref: sserve comp.os.386bsd.apps:1057 comp.os.linux.misc:11131 Path: sserve!newshost.anu.edu.au!munnari.oz.au!spool.mu.edu!agate!howland.reston.ans.net!math.ohio-state.edu!sol.ctr.columbia.edu!hamblin.math.byu.edu!news.byu.edu!cwis.isu.edu!u.cc.utah.edu!cs.weber.edu!terry From: terry@cs.weber.edu (Terry Lambert) Newsgroups: comp.os.386bsd.apps,comp.os.linux.misc Subject: Re: DOOM for X Date: 18 Mar 1994 02:06:35 GMT Organization: Weber State University, Ogden, UT Lines: 70 Distribution: inet Message-ID: <2mb2bb$8m@u.cc.utah.edu> References: <SJA.94Mar12174713@gamma.hut.fi> <2m0h7a$3ck@u.cc.utah.edu> <2m814r$bnp@news.mcs.kent.edu> NNTP-Posting-Host: cs.weber.edu In article <2m814r$bnp@news.mcs.kent.edu> borsburn@mcs.kent.edu (Bret Orsburn) writes: [ ... Me saying "Window managers should be in the X terminal" ... ] >But if you think about it for two minutes instead of one, you might >conclude otherwise. > >Whatever else it may be, a window manager is an application program. > >An application program needs an application run-time environment, and >for an X application that environment is probably going to be UNIX. > >So, to provide a general solution for local clients, your X Terminal >has to provide a UNIX application run-time environment (including all >of the application libraries your local clients might need). Well, first off, the only thing I'm really interested in is moving the window manager to the X terminal ... but as far as application environment, that's what xrdb an XInternAtom() are for. As far as API environment is concerned, a clock, a window manager, a print server, and an X telnet or rlogin or CTERM (DECNet) window all consume only those interesting resources that must be there for an X terminal to be an X terminal (ie: networking, display, and mouse/keyboard input services). For print services, I guess you'd need a local printer port (check the back of NCD, NCR, or GraphOn X terminals lately? Even a Wyse-50 has a printer port...). Second, the real benefit to RPC'ing the widgets is the reduction in wire traffic *on the average*. This is the same benefit I expect from a local window manager, which no longer has to get it's events from the X terminal over the wire and send events to the reparent clients (bouncing them through the X terminal), such as geometry management, etc. The problem is wire traffic; an 8 bit address range logical subnet on a 10MBit/S wire is simply too many hosts for an exclusively X traffic wire; putting the window manager on board divides the traffic by a factor of 2-3, depending on what the users are doing to the terminal. Finally, you are arguing from the specific to the general, which is logically invalid in any case. >That ratchets your system design a large notch closer to being a workstation. >(In fact, the only thing that distinguishes this design from a workstation >is the compromises you make to a "real" UNIX environment to save costs. >Those are the same compromises that all of those enlightened programmers >are going to come back and tell you about later.) This is the argument Sun tried to use (an failed at). Sun is now selling workstations running X server software under the name "X terminal". Sun wasn't very successful selling that world view, and they had a marketing department being paid big $ trying to back their story up. 8-). >You may just have designed away the cost/performance advantage you had >by building an X terminal instead of a workstation. Actually, you did the designing away when you offered to expand the environment *far* beyond what I needed to accomplish my goals. >And, just to make things interesting, you just designed an open system >instead of a closed system. I hope you're prepared to go into the programming >support business. Well, NCR did the designing, but given what AT&T now charges for the beasts, they can afford to support a couple of programmers here and there. 8-). Terry Lambert terry@cs.weber.edu --- Any opinions in this posting are my own and not those of my present or previous employers.