bzs@BU-CS.BU.EDU (Barry Shein) (10/25/87)
Re: Greg Earle's discussion provocater about the information/software explosion threatening to sink us all. I realize my Subject: was provocative and don't really mean to say that the "small is beautiful" crowd is obsolete, just that no one seems to be listening to them anymore. As Greg points out things like window and document systems (X, NeWS, TeX, InterLeaf et al) are behemoths of complexity and it doesn't look like anyone is offering any choice in the matter. By `choice', of course, I mean alternatives not just flames like "they're big, they stink!" which isn't really much of an answer except perhaps from god (pandora's box and all that, can't just tell the users to go back to dumb terminals and Model 34 teletypes, at least not without dabbling in demagoguery.) My first (and sustained) reaction to TeX, for example, was Blecch (as in "rhymes with".) Most fanatics of course view only the use of the system itself. My systems staff and I got to put it up and view with wonderment a 10,000 line single WEB (what's that?) module become a 15,000 line (something like that) single Pascal module. Then get undump to work on every system that people were screaming for it on, then suffer the various utilities (the font managers, the fonts, metafont, pxl fonts, pk, gf, dvi, bibtex, latex, slitex, lplain, detex, dvi2ps, amstex, magsteps, computer modern, etc, etc, etc.) I decided I knew a house of cards when I saw one and this would not be where *my* papers were going, even if it prevented me from having the most wonderful sigmas ever seen. It just wasn't that important. The current apparent weaknesses of Tex in previewing (yes, I've seen the systems, they all look very silly to me) and postscript standards (sure, some people are getting along, but I don't get the impression that postscript and TeX mesh very well, not at all sure why, font philosophy I assume is one reason) seem to be showing some current problems in TeX's ability to move into the present. I'm not surprised, it is typical of such behemoths of complexity, they are predictably lead-footed. In an attempt to appeal as some sort of standard they define everything in terms of the current state-of-the-art and cannot deliver on this rarified clarity as the world around them changes. Although I am picking on TeX (justified, in my opinion) I don't think it stands alone. As I said, it's just typical of such arrogance. To prognosticate I believe two things will happen: 1. People will appear who have better vision as to the underlying principles (it just *seems* like placing characters on a page shouldn't take several multi-hundred page volumes to explain, for example.) From this entire new approaches will obsolete the old. We are just in a rococo stage of affairs with these technologies which is normal in scientific progressions (first simple attempts, then 'great works' attempting to cover all ground by detailed case, then underlying principles are discovered and the field is reduced to a few fundamental tenets.) 2. We will see more use of "information appliances" as the Mac claimed to be. Turnkey packages will take care of these relatively mundane matters (eg. getting the kerning right.) People will tend to use tools which match their jobs and be satisfied with the lack of micro-creativity in return for some stability and simplicity. With any luck people will get back to the original goal (saying something useful rather than spending all one's effort making it look pretty, needless to say I've had it with font weenies.) Well, that should be enough for one sitting. -Barry Shein, Boston University
roy@alanine.phri (Roy Smith) (10/25/87)
Warning, this is one of my "keep typing until you get tired" messages. It's long, and it wanders, but I think it hits a few good points, so keep reading (or, at least, keep skimming). First, a digression on a point of only marginally-relevant trivia. "They" have been arguing in j-random newsgroup for the past week or so about the "right" was to represent a paragraph and/or sentence break in ascii text, so smart display software can reformat the paragraphs to fit the reader's screen. At first glance, I wondered why these people were getting so worked up. Then, when I attempted to edit out a sentence of the following quotation of Barry's words I discovered that emacs's idea of a sentence break (two spaces after the ".") doesn't match Barry's (just one space) and killed a whole paragraph when I meant to kill a sentence (M-K). Maybe it's really worth worrying about after all. Anyway ... > As Greg points out things like window and document systems (X, NeWS, TeX, > InterLeaf et al) are behemoths of complexity and it doesn't look like > anyone is offering any choice in the matter. By "complexity", I assume you mean complex to write and maintain the software. Well, I just got my demo tape of FrameMaker. FM is an InterLeaf type program for Sun workstations. Frame Technologies (the purveyors of FM) claims that the source code is about 5% of the size of InterLeaf's source and that the entire software development project took only a single man-year. On the other hand: text data bss dec hex 204800 589824 0 794624 c2000 /usr/local/bin/tex 977068 128056 49568 1154692 119e84 /usr/local/bin/maker (tex included for comparison purposes; this is the output of the Unix "size" command on a Sun-3). Regardless of how small the source is, or how fast it was written, a Mbyte of text *is* a bit on the large side. > it just *seems* like placing characters on a page shouldn't take several > multi-hundred page volumes to explain, for example Agreed! I havn't seen the full FM documentation (they don't ship it with the demo tape) but all I needed to get going on some moderately complex stuff (3-column output, multi fonts, headers and footers) was to read through the not-very-dense 25-page demo document (in the flavor of teach-emacs) and start playing around. Of course, the real win was that I was able to import 90% of my Macintosh skills -- they tried hard to make it look as much like a Mac as they could on a Sun screen, and they were largely successful. It also helped that I know a bit about typesetting so the contents of the various menus all made sense. At the risk of sounding like a commercial, do yourself a favor and get the FrameMaker free demo tape. It may not be the cat's pajamas in the WP department, but it's pretty close, and at a small fraction of what InterLeaf is charging (just under $1k per workstation to educational sites). It's not as fast as I might like, but considering what it does, it's really quite reasonable, after you get past the 2-minute startup time. I'm not sure I'm actually going to shell out the bucks for the real thing, but I'm impressed. > We will see more use of "information appliances" as the Mac claimed to > be. You sound like you're not convinced that they are. Why not? After all these years of teaching "dumb scientists" how to fight with Unix, it's really amazing to see people walk up to the Mac and start doing useful work right away. If that's not an "information appliance", I don't know what is. Also, why the past tense; "claimed to be"? It's not dead yet. Another digression: It's also really amazing to watch them all walk away with hot copies of everything on my hard disk. How do I convince people in this information age that theft of software is indeed theft. These otherwise honest, hard-working, ethical people practically laugh in my face when I patiently explain over and over again about copyrighted, but not copy-protected, software. The same guy who wouldn't dream of making a phone call using a fraudulent credit card number is perfectly happy to walk off with a $300 program. Not just that, but they even give me copies of stuff I don't want -- I can't tell you how many copies of Microsoft Word I've removed from my hard disk (it's a public-access machine). Sometimes I feel like a preacher in a whorehouse. Am I the only schmuck left in the world who actually respects copyrights? On a somewhat related note, what's the future of per-CPU software licenses. It's already confusing enough trying to decide what's a CPU and what's a terminal. The general opinion seems to be that my 3/50 is a CPU, even though it doesn't have a disk. FM enforces per-CPU licensing by doing some encryption magic on the host's internet address. When I explained this to somebody, they immediately wanted to know why we couldn't just license FM on a single Sun and have people rlogin to that host to run it. I explained that suntools doesn't run across rlogin connections, but it quickly occurred to me that NeWS and X do. Assuming that Suntools will slowly slide into oblivion to be replaced by NeWS and/or X, software developers are going to start to rethink yet-again the licensing question. How do you deal with software licensing when you're running an application on one host, a NeWS client on another, reading in the application binary from a NFS file system on a third, swapping it out to a ND partition on a fourth, and working with a document file stored on yet a fifth CPU? Maybe some of those machines are multi-CPU processors; maybe some have things like Sun's IPC card. Maybe I'm running SUPDUP in my comm controller to take the editing load off "the CPU". Confusing, very confusing. > People will tend to use tools which match their jobs and be satisfied > with the lack of micro-creativity in return for some stability and > simplicity. It's interesting -- you should see the pecking order we have around here. Person X is perfectly happy to produce mono-spaced text in whatever font happens to be the default, and write in the equations by hand. That drives me nuts, so I diddle with troff/eqn/tbl/bib and do it "right" (I consistantly get nicer-looking output using troff then he does using a Mac, and in less time). On the other hand, the graphics-artist who designed our Annual Report format is constantly getting on my case about "stupid details" like making the page numbers 12 pt instead of 11, and getting the header rules to be a Pica longer or shorter. Everybody has their own definition of where you draw the line between making it look nice and getting it out the door. Enought ramblings. Comments? s
tanner@ki4pv.UUCP (10/26/87)
Well, that note touched a sensitive location here. I used to use "roff" when I was in school, because everything it did was reduced to one page of instructions. I further reduced it to one macro and occasional use of ".ul 1" and ".ce 1". Well, I'm not using an 11 any more, and "roff" is now darkest history. I learned to use "nroff", which was still fairly simple if you weren't trying to do a whole lot. Creeping featurism is here, however, and I now maintain a couple of messy files of nroff/troff macros (yup, we're now using "troff" here for the final print). I pretend to understand traps, fonts, number registers, basic units, page offsets, double-column output, and all that happy stuff now. Of course, the telling point is that it takes longer to get a (prettier) document out the door now than it used to. Tanner Andrews, Systems CompuData, Inc. DeLand