ast@cs.vu.nl (Andy Tanenbaum) (12/16/88)
In article <575@uvicctr.UUCP> tholm@uvicctr.UUCP (Terrence W. Holm) writes: >This posting includes Douglas Gwyn's POSIX-compatible directory routines. Thanks a lot. I guess it is time to say a few words about the future of MINIX. I am now rewriting my architecture book. Suppose all goes well and I get it done by Sept 1, 1989. My next project will be to try and convert MINIX to be POSIX compatible, and then revise the book to reflect this new version. Since there will be a second edition of the book, I'll call it MINIX 2.0. My intention is to (a) try to be reasonably compatible with POSIX (b) change as little code as possible (c) keep the system as small as possible This means that when the POSIX standard says "shall" I'll seriously consider doing it. When it says "may" I won't. Thus no job control and other optional features. I'll also try to make the include directories and header files POSIX compatible, to allow people to port UNIX software more easily. In the original version I didn't bother much about making the include stuff compatible on the assumption that MINIX would never actually be used by anyone. It was intended for instructional use only, like Pascal. I am certainly interested in getting help. I'll probably have to tackle the kernel myself, but there are various areas that people who want to help can work on. Some examples: 1. A comprehensive list of what has to be changed would be useful. This means comparing each existing system call with the POSIX version thereof and writing down what has to be changed. In a few cases this is not clear. For example, in POSIX, ioctl has been split up into a large number of separate library calls. It is up to the implementer to figure out how many actual system calls there should be. 2. Many utility programs will have to be changed to use the new system calls. Any help in rewriting them will be welcome. Since Berkeley 4.4BSD will be POSIX compatible, people who have access to that can test modified MINIX software there. I am not sure what the release date is for 4.4. 3. New library routines will be needed and old ones will need fixing. 4. Test programs will be needed and the old ones will have to be patched up. 5. Etc, etc. If I end up doing all of this myself, it will take a few years. If other people help, it will go much faster. I think that if MINIX becomes POSIX compatible and both UNIX International (AT&T, SUN, and friends) and OSF (everyone else) are serious about conforming to POSIX, then MINIX will become a semi-serious system for people with PCs, ATs, 386s, and maybe Ataris. Especially if they want full source code. It might be a good idea for people planning to work on any of these areas to announce their intentions in the newsgroup, to look for partners and avoid duplication of effort. If the changes between 1.3 and 2.0 are not too huge, I would hope to post cdiff listings, so that all current users will be able to upgrade for free. In any event, P-H will certainly sell an upgrade kit, similar to the 1.x==>1.3 kit they will be selling in February 1989 for $30. Things like Stevie and Kermit might also be included in 2.0, assuming the owners of these programs agree. A decent text formatter would also be nice, if anyone can find one that is suitable both technically and in terms of ownership/licensing. Andy Tanenbaum (ast@cs.vu.nl)
wan@cory.Berkeley.EDU (Hank S. Wan) (12/16/88)
In article <1815@ast.cs.vu.nl> ast@cs.vu.nl (Andy Tanenbaum) writes: > > Since Berkeley 4.4BSD will be > POSIX compatible, people who have access to that can test modified MINIX > software there. I am not sure what the release date is for 4.4. I don't believe such a beast will ever exist. Supposedly the university no longer wants to support bug fixes so future releases, if any, will be from industry or other universities. -Hank Wan wan@cory.berkeley.edu
sater@cs.vu.nl (Hans van Staveren) (12/20/88)
In article <8359@pasteur.Berkeley.EDU> Hank S. Wan wrote: >In article <1815@ast.cs.vu.nl> Andy Tanenbaum writes: >> >> Since Berkeley 4.4BSD will be >> POSIX compatible, people who have access to that can test modified MINIX >> software there. I am not sure what the release date is for 4.4. > >I don't believe such a beast will ever exist. > >Supposedly the university no longer wants to support bug fixes so future >releases, if any, will be from industry or other universities. > >-Hank Wan > wan@cory.berkeley.edu At the NLUUG conference in the Netherlands on November 10, 1988 Keith Bostic told about the plans for BSD 4.4. He was a bit vague about release dates, but my impression was that it would again be distributed, and that 100% POSIX compliance would be one of the new improved things. This conference was dedicated to standardization, and Keith said they had some laughs about people from Berkeley now being invited to talk about standards. He seemed serious though. The best part of his talk was about wormhunting however, a subject that went down much better then all this dry standards work. Hans van Staveren Vrije Universiteit Amsterdam, Holland