bob@allosaur.cis.ohio-state.edu (Bob Sutterfield) (10/18/88)
(Note that followups have been redirected to comp.sys.next) In article <530@sactoh0.UUCP> sjrudek@sactoh0.UUCP (Steve J. Rudek) writes: >There are obviously a lot of USENET folks jumping up and down and >salivating because (a) NeXT is offering an "educational discount" and >(b) they are "students" (I surmise). Is it correct to assume that >*individual* students will also be eligible for this price discount? >(If so, I suppose I can always go back to school for a quarter:) That would have to depend upon the deal that NeXT has made with your local University computer store. There are certain restrictions and qualifications on just what sort of student can buy Macintoshes through OSU, for instance, though I don't know the exact terms. Besides that, they can't ramp up production fast enough to satisfy that kind of demand, so the cubes are being sold to carefully selected departments and projects that can make them shine. Individual students may have to wait. >I keep hearing that MACH is *derived* from UNIX -- and, in one >previous message, that the new release of the OS will be free from >AT&T licensing. This, to me, raises the obvious question: is MACH a >*complete* UNIX clone? Does it include basically *all* the utility >programs that come pretty much standard with AT&T/BSD UNIX? No, Mach is just 4.3BSD, with no SysVisms thrown in (that I know of). A Mach machine is one of the purest 4.3 systems you can find any more from a commercial UNIX producer, since they're all mixing in varying amounts of SysV in various ways. In fact, FSF says that the Mach group is working hard on removing any residual code that's still subject to the SysV license, so that FSF can use Mach in GNU. See the ongoing discussion in gnu.announce. >Just how (a) buggy and (b) compatible can we expect it to be? (a) They took CMU's Mach and "commercialized" it, removing a lot of the parochialisms found in CMU's distribution, like a dependence upon a printer named "third" because all the printers in Science Hall are on the third floor. They also cleaned up a lot of bugs. I don't know how much they're feeding back to the Mach group at CMU, but I understand they're in close communication, which will help all the other Mach port vendors as well (BBN, Encore, FSF, etc.) (b) If you had a VAX running Mach, you could run a 4.3 VAX binary on it. How much more compatible do you want? See the ongoing discussion in comp.sys.next. -=- Zippy sez, --Bob Are you still an ALCOHOLIC?
bob@allosaur.cis.ohio-state.edu (Bob Sutterfield) (10/18/88)
(Note that followups have been redirected to comp.sys.next) In article <1152@mmm.UUCP> schultz@mmm.UUCP (John C Schultz) writes: >Disk shuffling will still be a problem, escpecially if people >implement 100+ networks of NeXT machines, each of which need to be >backed up. Gaak! You're still thinking in personal computer terms, not workstation terms. Remember, this machine is for a University's networked workstation environnment. It's not intended for the home market! That's not snobbishness, it's a marketing choice. The reason the thing has no Winchester by default is (a) it makes more sense as a diskfree workstation talking over the network to a NFS file server (local / and /usr are *hard* to administer in the 100's); and (b) so that the appropriate amount of Winchester storage can be added in the case that someone needs to configure one as a server or the very occasional standalone. In this environment, the optifloppy is used for software distribution. If a NeXT machine is a server, it can be used for dump or rdump. Usually, though, dumps will be done to a central server's 1/2" tape drive. The optifloppy on an individual workstation in a lab will be used the same way students use magfloppies on Macintoshes now: to carry around their personal work in their backpacks from lab to lab. -=- Zippy sez, --Bob I just had a NOSE JOB!!
bob@allosaur.cis.ohio-state.edu (Bob Sutterfield) (10/18/88)
(Note that followups have been redirected to comp.sys.next) In article <34946@clyde.ATT.COM> wtr@moss.UUCP (Bill Rankin) writes: >[ note: assuming bare-bones next-box here. just purchased. nobody >[ yet to ether-up to. just the optical drive. Bad Assumption. The cube isn't being sold into such environments right now: all the first customers will in a workstation/server environment already, and you'll be chugging NFS from a Sun or a Pyramid or whatever in the Computer Science department. If you don't have a server handy, then you'll have a cube configured with a Winchester, possibly as a new server for a small cluster. This is how they'll go into the Dance department. -=- Zippy sez, --Bob I'm not an Iranian!! I voted for Dianne Feinstein!!
guy@auspex.UUCP (Guy Harris) (10/19/88)
>No, Mach is just 4.3BSD, with no SysVisms thrown in (that I know of). "Just 4.3BSD"? I wouldn't go that far at all. The kernel has a heck of a lot of non-4.3BSD code in it (and I think they want to move a lot of "kernel" code *out* of the kernel entirely). > A Mach machine is one of the purest 4.3 systems you can find any more > from a commercial UNIX producer, since they're all mixing in varying > amounts of SysV in various ways. Which isn't necessarily a good thing; "the varying amounts of SysV" aren't just thrown in because S5 compatibility is a checklist item, much of it is thrown in because it's good stuff (HDB UUCP, the S5 Bourne shell, the S5 "make", the S5 "old awk" which is faster than the earlier version that comes with 4.3BSD, the S5R3.1 "new awk", etc.). >In fact, FSF says that the Mach group is working hard on removing any >residual code that's still subject to the SysV license, Which includes a lot of 4.3BSD kernel code (which is presumably one reason they're starting with Mach). You could call Mach a "4.3BSD look-alike", except that 1) "look-alike" seems often to be somewhat pejorative, implying that it may not be the Real Thing and may not work like the original, and 2) a lot of the code comes from 4.3BSD (I suspect they didn't bother rewriting all of the C library or the utilities, although they may have changed some of the C library to make it work better with multiple threads in a single address space - I think they did change standard I/O so that it could use file mapping).
hedrick@athos.rutgers.edu (Charles Hedrick) (10/20/88)
> William Somsky argues that the machine would tend to be used >as a standalone PC. I argue that the machine would be used as a >networked workstation. A removable optical disk is the wrong medium for student lab use. If you've got file servers, you want the software to be stored there. If there's any removable medium at all, you want it to be something cheap like a floppy, for a student to put his own files on. Unless they provide some way to lock a given optical disk in the machine permanently, we sure can't put a system like that out in public.
dorn@fabscal.UUCP (Alan Dorn Hetzel) (10/20/88)
Re: software installation with 1 disk drive.
1) DEC has a similar problem with RC25 based Microvax-II systems,
the disk had a 25Mb fixed section, and a 25Mb removable section,
but they shared a common spindle. To load a removable disk, you
had to spin down the system disk as well. They fixed the problem
by teaching VMS to understand extended absence of the system disk.
It essentially just tapped its feet and waited for you to give it
the disk back.
2) Since the NeXT has 8Mb of main memory, it is probably practical to
have a disk copy/installation utility which pages all nonessential
code out, lock itself in memory, and then performs the copy using
all spare memory as a buffer. Agreed, this would really suck for
copying a full disk, but the average installation could probably be
done in 1-3 passes.
Opinions: ?
Dornjlemon@cory.Berkeley.EDU (Jonathan Lemon) (10/20/88)
In article <Oct.19.19.20.19.1988.5961@athos.rutgers.edu> hedrick@athos.rutgers.edu (Charles Hedrick) writes: >A removable optical disk is the wrong medium for student lab use. If >you've got file servers, you want the software to be stored there. If >there's any removable medium at all, you want it to be something cheap >like a floppy, for a student to put his own files on. Unless they >provide some way to lock a given optical disk in the machine >permanently, we sure can't put a system like that out in public. Why bother with floppies? What's wrong with the optical disk? I buy one optical disk at $50, and use it on the cube from then on out. Compare this to having to buy, oh, "The Red Dragon Book" at $40, and use it for only one semester before it goes in my 'reference heap' with the other books I have. -- Jonathan ...ucbvax!cory!jlemon
bzs@xenna (Barry Shein) (10/23/88)
From: hedrick@athos.rutgers.edu (Charles Hedrick) >A removable optical disk is the wrong medium for student lab use. If >you've got file servers, you want the software to be stored there. If >there's any removable medium at all, you want it to be something cheap >like a floppy, for a student to put his own files on. Unless they >provide some way to lock a given optical disk in the machine >permanently, we sure can't put a system like that out in public. Macintosh floppies typically have system folders on them, why not a basic unix dist on each user's optical disk (which can then go NFS mount a server directory for more software)? At least enough to boot up and get going with. The university basically doesn't own any of the optical disks that are going into the machines. The optical disks are $50 for 256MB, even with a pretty hefty Unix distribution (and swap/page) on them there's probably over 150MB for a student's files. I don't think $50 is expensive for 150MB of personal storage, in fact, compared with floppies it's downright cheap, I'd be surprised if many students could use that much in their entire undergrad career (how much disk do you give undergrads right now?) I could imagine the optical disks being sold pre-formatted with Unix on them, why not (how else?) So you walk up to a machine, plug it in, boot and go to work. I don't understand the objection, other than perhaps a feeling that a student walking around with an entire Unix binary distribution in his/her pocket seems wrong, why? The only possible objection I can see is that student's software will become outdated. Big deal, mac user's run into that all the time. In the first place, it's not necessarily a bad thing (ie. their applications they may have paid for/built will work with *their* version of the system) and besides, the optical disk could be updated, probably by buying a new one for $50 and transfering all the user files to it (probably more reliable than trying to update the user's disk in place), the old disk becomes a backup, an extra, or even gets traded in for credit on the new one and is recycled. How often would it be an issue? Once a year I would guess, $50 every September with no particular rush to update to it in most cases. -Barry Shein, ||Encore||