flamer@omsvax.UUCP (10/07/83)
I don't think anything is wrong with making improvements to Unix in its varied forms, but I do agree with one point: we need to make moves in the standardization direction. I noticed that the vi/ex editor showed up in the distribution of System V. This is a token movement of the type I suggest. But to make Unix survive as a legitimate OS, some people need to get together and make their own ANSI-like committee. I don't really think ANSI should get involved, because they generally take a long time to do things. The people of which I speak are BTL, BSD, Microsoft, etc., etc., all the people who are responsible for the various Un*ces. Maybe a big conference once every two years, where these people DECIDE on the contents of STANDARD Unix, undoubtedly being a melding of the various efforts, and all Un*x vendors abide by the decisions. Change and improvement are not bad, as long as we have a consistant, controlled effort. Jim Trethewey HF2-2-243 Intel Corporation Hillsboro, OR 97124 (503) 681-5444 ..!tektronix!ogcvax!omsvax!flamer ..!decvax!microsoft!omsvax!flamer
wapd@houxj.UUCP (Bill Dietrich) (10/10/83)
Aren't BTL's (or WECO's or ATT's) agreements with Motorola and Intel and someone else moves toward standardization ? As I understand it, those companies will write Unix versions for their machines to ATT's specifications (i.e. identical to System V, except running on the different hardware). Bill Dietrich houxj!wapd
ian@utcsstat.UUCP (Ian F. Darwin, Toronto, Canada <ian@utcsstat.uucp>) (10/10/83)
There is currently a UNIX standardisation effort well underway. The /usr/group UNIX standards committee has a standard for the system interface (mainly sections 2 and 3 of the manual set) which is in final draft stage at this time. The committee has, as you suggested, representation from major vendors (at least Fortune, Pyramid, Microsoft, ITT, Bell Labs, Plexus, DEC, CRDS, HP, RLG, RM, SUN, ITS, WHitesmiths, HCR, Mark WIlliams. UniSoft is the only major port-provider not on the list of names in the version of the Draft which I have). Their current work is on the "System Interface" standard, but they have plans for a C standard and a user interface standard eventually. The System Interface is the one of most concernt to those writing commercial applications, so that's where they went first.
jpl@mhtsa.UUCP (10/11/83)
I thought UNIX(tm) System V, distributed by Western Electric, was the standard UNIX(tm) system. Jeff Lankford mhtsa!jpl
gwyn%brl-vld@sri-unix.UUCP (10/14/83)
From: Doug Gwyn (VLD/VMB) <gwyn@brl-vld> The UNIX standardization problem does not go away by just "defining" a standard, if sufficiently many people have reason not to follow it. The fundamental problem is that there is NO single version of UNIX (that I am aware of) that doesn't have some deficiencies that other versions have fixed. Often the deficiencies were fixed differently in different flavors of UNIX. If there were one common version that provided reasonable support for everything an application programmer needed to be doing, we would not have such a proliferation of UNIXes. Unfortunately, UNIX developers have not always been responsive to real user requirements (look how long IPC was neglected), and different development groups have different interests and concerns. That is why I think there is a real need for dialogue among these persons, if UNIX is going to get out of this fragmentation rut.
gwyn%brl-vld@sri-unix.UUCP (10/19/83)
From: Doug Gwyn (VLD/VMB) <gwyn@brl-vld> I agree that the issue of UNIX standards is a very important one and that the DEVELOPERS of various UNIXes should meet on occasion to discuss the issues and to try to coordinate their directions. I believe that USENIX conferences years ago used to accomplish a lot technically, but in recent years the technical content has dropped significantly. How I miss the "guru panels" where internals could be freely discussed! How about setting up a DEVELOPER Birds-of-a-Feather session at the Washington USENIX conference? It would perhaps be necessary to show proof of source license, although I think that could be avoided. I would particularly like to see the following categories of people present in a panel at such a session: (a) Bell Labs Research people (b) Berkeley Software Distribution developers (c) Western Electric developers (d) Well-known gurus (not to mention any names here, since I would miss somebody important, but I do have about a dozen "old-timers" in mind). If there is sufficient interest from the right people, I would gladly work on setting up this "session".