krowitz@mit-kermit.UUCP (David Krowitz) (05/28/87)
I agree. Pure unix is obsolete. Even the venerable Decsystem 10 running TOPS-10 (and not the newer BBN Tenex, or TOPS-20) had better file protection scheme than BSD4.2 does. The Apollo distributed registries are *much* better than /etc/passwd and /etc/group. I can manage 15 Apollo nodes belonging to seven professors with (roughly) sixty graduate students with less work than it takes to manage one Sun 3 and one Alliant FX/1. It is my ferverent hope that in becoming more 'real unix' like that Apollo does not give up the advantages they already have over BSD4.2 (like commands that have a consistant set of switches, not some with a '-' and some without; typed objects which allow my programs to test if they are reading the right kind of file before they start reading garbage; an extensible file system which I can use to create mu own file types and device drivers; and a set of disk utilities (INVOL, SALVOL, etc.) which don't require me to deal with the number of tracks, heads, sectors per tracks, and bytes per block of the disk every time I want to change the size of a partition). -- David Krowitz mit-erl!mit-kermit!krowitz@eddie.mit.edu mit-erl!mit-kermit!krowitz@mit-eddie.arpa krowitz@mit-mc.arpa (in order of decreasing preference)
mkhaw@teknowledge-vaxc.UUCP (05/29/87)
I think Apollos are fine machines for running Aegis, and I have few quibbles with Aegis: the commands are just enough like Unix to be irritating to learn and confusing when I switch between an Aegis pad and a Domain/IX pad; "pst" doesn't identify Domain/IX processes in useful ways (shows some unusable uid number something or other); commands like "dlf" don't work well with certain Unix filenames like ".cshrc" (i.e., ":.cshrc" in Aegis); and if you telnet to a Domain/IX machine on a terminal and then try to "/com/login" you get silly things like the message "Not any password!", after which Aegis merrily echoes your password on the screen as you type it in. With regard to the original question as to whether or not an Apollo is a Unix box: I don't think Domain/IX, so far, is real Unix in the sense that you have to know both Unix and Aegis to administer a Domain/IX system, and you can't have Unix on an Apollo without Aegis. Aegis and the Apollo Domain ring may be technically superior to Unix and tcp/ip ethernet, but getting locked into a proprietary OS and networking hardware, and learning a user-environment (i.e., a set of commands) that you can't use on other machines is a lose. Now, if Domain/IX were a "native" Apollo Unix with extensions like acls, etc., *that* would be something! Mike Khaw -- internet: mkhaw@teknowledge-vaxc.arpa usenet: {hplabs|sun|ucbvax|decwrl|sri-unix}!mkhaw%teknowledge-vaxc.arpa USnail: Teknowledge Inc, 1850 Embarcadero Rd, POB 10119, Palo Alto, CA 94303
lid@cernvax.UUCP (05/29/87)
I agree, Unix is obsolete, I'm managing here a 34 Apollo nodes split in two rings bridged toghether and I still have spare time. Everything just works fine (or almost fine) and I wouldn't really like to go to Unix, this wouldn't solve any of the problems we have now with workstations. Achille Petrilli
benoni@ssc-vax.UUCP (Charles L Ditzel) (05/31/87)
In article <486@cernvax.UUCP>, lid@cernvax.UUCP (lid) writes: > > I agree, Unix is obsolete, I'm managing here a 34 Apollo nodes split in two > rings bridged toghether and I still have spare time. Everything just works fine > (or almost fine) and I wouldn't really like to go to Unix, this wouldn't solve > any of the problems we have now with workstations. > > Achille Petrilli Right!! Unix is obsolete because a a couple of sites run Aegis and they like it. Somehow this misses the larger point that the *industry* is going to Unix. I don't know if you have noticed but a lot of workstation vendors are taking pains to put together Unix boxes. Proprietary systems such as Aegis are increasingly frowned on. I think Apollo has done somethings very nicely but on the larger issue of Unix it has been not only very slow but very stubborn. Sun seems to have taken advantage of Apollo's lack of vision with regard to where the industry is going. What one of my Apollo instructors demeaningly called a "rinky-dink company" is outselling Apollo precisely because Apollo's *are* *not* Unix enough. (Of course there are some other excellent reasons why Suns are outselling Apollos - like a low-end diskless, high resolution monochrome workstation that runs for $4K.) I do not like ACLs, i am not convinced they have any value given that I have to be compatible with Unix Vaxes, PC RTs, Masscomps, Suns, etc. Use of ACL system calls are bound to fail on other Unix systems. Furthermore, the mapping between ACLs are Unix permissions are flawed. Another problem is that the acl_cache sometimes gets corrupted and you wind up with ownerships of '-1' or some such nonsense. Of course there is another kludge for that, some utilities that restore the system permission to the ACL setup. Despite some good ideas that may eventually find there way into Unix, Aegis is dead due to industry treads, it's proprietary nature and Unix.