ian@utcsstat.UUCP (Ian F. Darwin) (02/05/84)
Who's next? Really, who is now? I agree with the argument that UCSD-Pascal was just another fad (along with Forth). CPM is here to stay. It has the advantage of simplicity. I'm not sure I agree about the ``simplicity'' of CP/M. See below. Everyone knows CPM. Yes, unfortunately. The retarding effect this has on serious development will likely never be fully realised. I also agree that UNIX will *NOT* be the next major O.S. for micro's. In order for something to beat out CPM, it has to have more advantages than dissadvantages. For the novice user, UNIX just plain stinks. It is hard to use. It is cryptic. It has horrible error messages. I disagree that UNIX ``just plain stinks'' for ``novice users''. A novice user is someone who is in process of becoming an experienced user. After a few hours, CPM becomes boring and *very* limiting. You can learn the same simple tasks on UNIX in a few hours, but then there is a lot more that you can learn to do with UNIX that isn't available with CP/M. And, as has already been mentioned, which is more cryptic: cp oldfile newfile or PIP NEWFIL.E=OLDFIL.E ? Which more horribly vague: cp: write error or BAD SECTOR And lastly [it] is much too easy to do horrible damage to your files without even trying (very hard). Like rm *? But CP/M lets you era * or *.* just as easily. By redirecting output onto your files? Aha, that is just the measure of how easy it is to create files. And if you really are a keyboard klutz, the c-shell (Berkeley's attempt at writing a better shell) lets you `set noclobber' to remove some of the easy ways of overwriting files. In any event, important files can be made non-writable in both systems. Further (as has also been mentioned several times before) [UNIX] requires excessive amounts of central memory and disc storage. What do you mean, excessive? You mean you can't run in 16KB? Well, that's true but given the price of memory these days, I say: so what? Nobody is his/her right mind buys 64KB memory boards anymore, since 256KB ones only cost a bit more, and many UNIXes can run with that. Hard disk? Nobody serious about program development will buy a floppy- only system these days. Run-time systems definitely. Or even stand-alone systems with only serial I/O and external monitoring. But even Digital Research long ago gave up on CP/M for development, and (my rumor mill tells me) bought a VAX on which they cross-compile and cross-debug. I had much rather develop real-time stuff under UNIX, using the UNIX development tools, and *then* recompile it to run stand-alone or on a floppy-based machine, than ever try to develop real code on a toy system. I know whereof I speak. My home system can be switched between CP/M and UNIX just by exchanging CPU board and powering on the floppy and loading the CP/M boot disk. It's easy, and can be done in under 3 minutes each way. Yet I have not gone down to C(PM) level for a long time. The reason? UNIX is (despite your claims) a much friendlier system. And a much more productive environment. And the only complete operating system under discussion. Lastly, the multi-tasking features of UNIX (at least all the implementations I have inventigated) are buried so far down into the KERNEL that user programs can not use them. (yeah, I know you can pipe things all over the place and spawn processes, but get them to talk to each other in real time.) Let's keep some basic terminology straight. Multi-tasking (makeing multiple processes) is easy to do in UNIX, either from the keyboard or from within a program. Real-time stuff (loosely meaning: very fast response to external signals such as A/D/A converters, signal counters, etc., etc., can and has been done (see Proceedings from most USENIX technical conferences). If you feel you must run your real-time data acquisition stuff stand-alone, see above; develop under a real operating system, run real-time stuff stand-alone. Why should anyone WANT to use UNIX (other than the 'fad' aspect of it)[?] Why would anyone want to deny themself the benefits of having an operating system? No, I don't believe that UNIX(tm) is the next popular O.S. for micros. If it weren't for the public abhoration for anything INTEL does until some outside vendor changes the name and sells it, I would say RMX-86 would stand a good chance. Maybe this is a reflection of how user-friendly their stuff really is. This does bring to light the two-sided nature of the next 'popular' O.S. First of all it will have to support multi-users/multi-tasks. UNIX does. Secondly, it will have to *LOOK* like CPM/MSDOS/... Now for all you UNIX(tm) fans out there, don't fret. UNIX can easily be made to look like CP/M, just by making a few aliases and only using a small subset of it (or by buying one of several commercial CP/M emulators for UNIX). CP/M can never be made to look like full UNIX, although ``shell'' (CCP) replacements can give it something of the flavor of a UNIX subset. It will also have to support a UNIX(tm) shell. What is this fantastic system going to be? It beats the heck out of me. Everyone I have talked to is either playing 'catch up' or is trying to band-aid currently existing systems to do more and more. For once I agree with you, although I suspect that we are using terms in somewhat different ways. It does worry me that the entire effort of Silicon Valley seems to be directed to producing products that are 0.05 % better, and 0.01 % cheaper, than the competition. The Valley fiddles while Rome burns... and the sun is rising in Japan, folks. Little effort has been given to taking one of the multi-tasking packages and build reasonable support utilities over it. (You pick it, VRTX or MTOS are supported on several machines including the 68k and '86). Now that's a bigger undertaking than you realise. And UNIX already has it! Why bother? Do I sense the presence of the ``not invented here'' disease? The only integrated package for a micro that I know that does all this is RMX-86, and it is an INTEL product. It also only runs on an 8086 and (here goes again) is written in PLM-86. Well, you're entitled to believe what you like about RMX. So what is the answer? Don't ask me, I only have the questions. For this generation of computerists, UNIX is the only reasonable answer. For the future, only time and the marketplace will tell. Mike Spann Ian F. Darwin, Toronto uucp: utcsstat!darwin!ian -- Ian F. Darwin, Toronto uucp: utcsstat!ian
richard@sequent.UUCP (02/11/84)
I left Victor Technologies (RIP) a while back after spending quite some time using first CP/M and then MS-DOS. I was in the Advanced Development group, so I really know those systems. I joined a company doing development on 4.2BSD. Recently I turned on my Victor at home, and sat there feeling crippled! There was so much that simply *couldn't* be done with MS-DOS, it wasn't funny. Having used csh on 4.2, my mind is made up. There's no question as to which system is easier to use or more powerful. For novices, they're both pretty much dogs. But does CP/M or MS-DOS have on-line manuals? How about Learn? And what's this stuff with "A>" and "B>"? Admittedly, for unix to become the accepted system it will be, it'll have to have alot of user-friendlyness added. But at least with unix, all you have to do is build a business-user oriented shell. Or a word-processor oriented shell. (There's a group creating a menu-driven shell - don't remember details). The system is so adaptable, it will fit all these bills. (Startup idea?: build shell for non-user types?) Eventually, we'll all be playing with dynabooks with mindlinks, thoughtputs, biomolecular circuitry (that's powered of ambient heat differentials) and learning-expert systems. But unix increases productivity from the simpler systems, and that's the key. from the confused and bleeding fingertips of ...!sequent!richard