ergo@netcom.UUCP (Isaac Rabinovitch) (09/03/90)
In <1990Sep2.093805.13164@NCoast.ORG> ac119@cleveland.freenet.edu writes: >Ad for Coherent mentioned that its users will be able to access news and files >via UUCP. This is a product I definitely want to learn more about. >-- >NEIL PARKS I've had a similar thought. I have access to a pretty good "public access" Unix system, so I should be satisfied, but doing my interaction over a 200 cps modem connection (typically several seconds to redraw the screen) and interacting with zillion-moded Unix software gets old. And eventually I'll probably have to start paying for the phone-line-time I'm using. So being able to do my news and mail interaction on my PC makes sense. But I don't think Coherent is the way to go for this. It and similar alternatives (some other form of Unix for the PC; news/mail software that runs under MS-DOS; buying an orphaned Unix micro, which might actually be the cheapest way to go!) all amount to the same thing: starting your own network node. That means administrative headaches, but worst of all it means "downloading" every message/posting that I *might* want to see. What I want is to preview/select messages on a host system, and download only the ones that interest me. As I see it, the big problem is that Unix software is still living in 1978, when almost everybody used a terminal, a special-purpose machine with little smarts beyond cursor addressing and efficient screen updating. Now, we're all using full fledged computer systems that generally have hard disks and a ton of RAM. But what do we do when we want to talk to Unix (or BBS, or most dialup) systems? We run an emulator that does little more than make that fancy 1990 computer pretend it's a 1978 terminal. Oh, we've got file transfers and keyboard macros, but those are minor features tacked onto that old Jimmy-Carter-era technology. Why should a Unix system assume my system can only do one thing at a time? Right now, I read news by browsing a list of messages, *one* screen at a time, picking the ones I want to see, whereupon the newsreader (nn in my case) sends them two me, *one* screen at a time. What the host system should do is send my sytem at *computer* speeds, not human speeds. While I'm reading the information in one screen, the two systems, instead of twiddling their thumbs, are "pipelining" the other information I'll probably want to see. This requires: clever new transmission protocols; clever new application design; multitasking -- but all of it's *quite* feasible if anybody's got the skill (no problem there) and imagination (well...) Hey, you Windows and Desqview types? You listening? If netnews and mail don't interest you, think about how must "faster" Prodigy would be if screens were pipelined, instead of waiting for a specific command and 4 seconds of downloading... -- ergo@netcom.uucp Isaac Rabinovitch atina!pyramid!apple!netcom!ergo Silicon Valley, CA uunet!mimsy!ames!claris!netcom!ergo Disclaimer: I am what I am, and that's all what I am!
wnp@iiasa.AT (wolf paul) (09/04/90)
In article <13111@netcom.UUCP> ergo@netcom.uucp writes:
)interaction over a 200 cps modem connection (typically several seconds
...
)Why should a Unix system assume my system can only do one thing at a
)time? Right now, I read news by browsing a list of messages, *one*
)screen at a time, picking the ones I want to see, whereupon the
)newsreader (nn in my case) sends them two me, *one* screen at a time.
)What the host system should do is send my sytem at *computer* speeds,
)not human speeds. While I'm reading the information in one screen,
)the two systems, instead of twiddling their thumbs, are "pipelining"
)the other information I'll probably want to see.
)
)This requires: clever new transmission protocols; clever new
)application design; multitasking -- but all of it's *quite* feasible
)if anybody's got the skill (no problem there) and imagination
)(well...)
Ah, but if it's slow READING news over a 200 cps modem connection,
how slow do you think if would be if a fast file transfer took
place over the same modem connection while you were reading the next
article or even just header?
--
Wolf N. Paul, IIASA, A - 2361 Laxenburg, Austria, Europe
PHONE: +43-2236-71521-465 FAX: +43-2236-71313 UUCP: uunet!iiasa.at!wnp
INTERNET: wnp%iiasa.at@uunet.uu.net BITNET: tuvie!iiasa!wnp@awiuni01.BITNET
* * * * Kurt Waldheim for President (of Mars, of course!) * * * *woods@eci386.uucp (Greg A. Woods) (09/05/90)
In article <13111@netcom.UUCP> ergo@netcom.uucp writes: > As I see it, the big problem is that Unix software is still living in > 1978, when almost everybody used a terminal, a special-purpose machine > with little smarts beyond cursor addressing and efficient screen updating. I certainly wouldn't go that far. In fact, you'd better be awful careful about just what you call "UNIX" software. > Why should a Unix system assume my system can only do one thing at a > time? Right now, I read news by browsing a list of messages, *one* > screen at a time, picking the ones I want to see, whereupon the > newsreader (nn in my case) sends them two me, *one* screen at a time. It doesn't. > This requires: clever new transmission protocols; clever new > application design; multitasking -- but all of it's *quite* feasible > if anybody's got the skill (no problem there) and imagination > (well...) NO, it certainly doesn't. There are many current standard or de-facto standard protocols and designs currently in use: The X-Window System - many versions for various "PC"'s. Layers - no available versions for any "PC"'s. Various other products - mostly for Mac's. There are X implementations for several "PC"'s, such as IBM-PC's Mac's, etc. I believe some even support SLIP, and would work reasonably well with high speed modems (such as Telebit's). For local workstations, high speed LAN's can be used. However, the best solution, especially for those limited by slower communications lines, is layers, also known as the SysV Windowing System. This very elegant system has been in use by AT&T on various "smart" terminals (jerq, blit, 5620-DMD, 615, 630, 705, 730, etc.) for quite a few years. It comes standard (as an operating environment) on all true System V ports (some vendors, such as ISC did not provide layers). Unfortunately, it doesn't appear that the IBM-PC version of layers ever made it out of the Labs. Never-the-less, such an implementation should not be too hard for MS-DOS, or the Mac, Amiga, Atari, etc. With a full "PC" at your disposal, as with some of the smarter terminals mentioned above, a full development system could be provided, such that layers applications could be ported to such an implementation. However, unlike with the terminals, a PC, with a built-in disk drive, would not need to have such applications stored on the host and downloaded to the PC. In fact development of such applications could be done entirely on the PC, again unlike on the terminals where the host must provide a complete development system. [However, I would still prefer to do development work on a UNIX system, where all the proper tools are available in an integrated environment.] -- Greg A. Woods woods@{eci386,gate,robohack,ontmoh,tmsoft}.UUCP +1-416-443-1734 [h] +1-416-595-5425 [w] VE3-TCP Toronto, Ontario CANADA
les@chinet.chi.il.us (Leslie Mikesell) (09/05/90)
In article <878@iiasa.UUCP> wnp@iiasa.UUCP (wolf paul) writes: >Ah, but if it's slow READING news over a 200 cps modem connection, >how slow do you think if would be if a fast file transfer took >place over the same modem connection while you were reading the next >article or even just header? The trick would be to decouple them completely and use data compression during the transfer. First pick an article size that can be transfered in slightly more than the amount of time that it would take you to decide if you wanted it if you looked at the headers first. Send anything smaller over without asking, plus the headers and a small portion of the larger articles. If you read an incomplete large article and want the rest of it, a query would be passed back and it would be injected into the transfer in progress. A provision would have to be made for batching the queries, though, because unless you read very fast or have a very slow modem, the articles are going to arrive faster then you can read them so you will want to disconnect for part of the reading time. For groups where you only read a small percentage of the articles, you could send only a few header lines (perhaps Subject, From, and Lines) automatically, and then your reader would allow you to mark the ones to retrieve (on the next connection if you choose to disconnect whenever possible). Les Mikesell les@chinet.chi.il.us
ken@dali.gatech.edu (Ken Seefried iii) (09/05/90)
In article <1990Sep4.172707.434@eci386.uucp> woods@eci386.UUCP (Greg A. Woods) writes: > >However, the best solution, especially for those limited by slower >communications lines, is layers, also known as the SysV Windowing >System. This very elegant system has been in use by AT&T on various >"smart" terminals (jerq, blit, 5620-DMD, 615, 630, 705, 730, etc.) for >quite a few years. It comes standard (as an operating environment) on >all true System V ports (some vendors, such as ISC did not provide >layers). Unfortunately, it doesn't appear that the IBM-PC version of >layers ever made it out of the Labs. > Watch telling this to a Georgia Tech person. When I got here (circa 1985), this place was flooded with 5620's. When they weren't dropping like flies from hardware problems, they were little more than a big pain in the butt to programme (and I do X for a living). Robert Vidua did some interesting things with them, but few others did (and we did try). The only bonus to the Georgia Tech user community was the fact that you could put a whole page on the display...I'd buy one just for this. Don't get me wrong...I thought they were a great idea (thats why I tried to learn how to programme them long after they had been phased out for the general users), but they just didn't live up to the promise... P.S. - Rob Pike's BLIT papers are a must read for anyone doing windowing system design. He's a little radical, but he has some good points... -- ken seefried iii ken@dali.gatech.edu "Vee haf veyz off making you talk...release da veasles..."
meadb@boulder.Colorado.EDU (MEAD BENNETT ROSS) (09/06/90)
In article <1990Sep4.172707.434@eci386.uucp> woods@eci386.UUCP (Greg A. Woods) writes: >NO, it certainly doesn't. [require new and innovative tools and software]. >There are many current standard or de-facto >standard protocols and designs currently in use: > > The X-Window System - many versions for various "PC"'s. > Layers - no available versions for any "PC"'s. > Various other products - mostly for Mac's. There is a pd program called screen posted to c.s.u a couple of years ago that creates layers that you can flip through. It's very handy when your news server is really slow: I sometimes go of and do little administrative tasks when it gets too slow. Is this program like layers, and is it portable to a pc with UN*X? OK so you have to do a little fiddling when screen mis- behaves, but on the whole its a great program. | | meadb%tramp@boulder.colorado.edu | | | | | | | | | | | | | | | | | meadb!tramp!boulder!ncar.... | | | | | | | | | | | | | | | | |meadb!tramp!boulder!sunybcs!rutgers...| | | | | | | | | |