rbk@sequel.UUCP (10/02/83)
I received a number of requests for whatever I found out about the Newcastle Connection. I got a number of responses; the following were typical: ********** An article on The Newcastle Connection can be found in Software -- Practice and Experience, Vol 12, 1147-1162, 1982. I have seen the system in operation at U. of Newc., and we have ordered a number of copies for our lab. While it can be criticized for implementation choices, conceptually it is the cleanest approach I have seen. ********** Look in the December 1982 issue of Software--Practice and Experience, for an article by Brownbridge, Marshall, and Randell entitled: "The Newcastle Connection, or, UNIXes of the World Unite". (pages 1147-1162). ********** That's right, the folks at Newcastle were able to do everything with library routines -- they didn't want to make any kernel changes! They used a Cambridge ring with a bunch of LSI-11/23's as their processors. You can read an article on it in, 'Software-Practice and Experience', Vol. 12 (pg. 1147-1162), 1982. ********** The original paper on the Newcastle system is: "The Newcastle connection *or* UNIXes of the world unite!" D. R. Brownbridge, L. F. Marshall, B. Randell Published December 1982 in Software - practice and experience Vol. 12 no. 12, pages 1147-1162. Perhaps I ought to make one comment about it. It requires recompilation of utilities in order to take advantage of it. Thus you need the source, even if you don't need to change it. I believe that it achieves its objectives by subverting the C library - don't hope for much with assembler progs, e.g. the assembler, or a Basic interpreter. ********** No, it's not better than Suns (you mean berkeley, actually) and the fact that it has all this user code in it, is going to be quite a pain!!. also there is kernel support, but it's slower than 4.2bsd ********** I heard a gentleman from Newcastle on Tyne at the USENIX BOF on distributed systems. The Newcaslte connection is implemented with PDP-11/23's with a Cambridge Ring Network. The software is all in libraries, since they didn't have a source license to play with. My impression was that you wouldn't want it as a commercial product. You can't enforce accounting or security from libraries. ********** The (many times) referened paper was obtained, and I've read it. The approach does seem interesting, but there's not much performance data presented in the paper. Of interesting note is that the positive comments outnumbered the negative ones. Be your own judge. -- Bob Beck Sequel Computer Systems ...ogcvax!sequel!rbk (503)627-9809