jkh@meepmeep.pcs.com (Jordan K. Hubbard) (06/09/90)
I, like many of the folks on this list (I assume), am interested in Bruce's minix port to the pc532 but also somewhat reluctant to jump into the morass of multiple patches, revisions and strange licensing issues without some assurance that I won't simply be buried by it all. I am also not adverse to shelling out some bucks to Prentice-Hall (since it's not big bucks) for release X.XX if I know for a fact that I can use it for this project, and that there's a reasonable upgrade path in sight. To this end, I think a few works from Bruce about eventual 1.5.X compliance (or whatever's being bandied around as the "latest" in comp.os.minix), and what should be purchaed from PH at this point would be very reassuring. Naturally, everybody wants the latest and greatest (if for no other reason than being able to apply patches as they come across the net) without screwing Prof. Tannenbaum out of the few bucks he probably makes from all this, so what do we do? Jordan
culberts@hplwbc.hpl.hp.com (Bruce Culbertson) (06/09/90)
In response to Jordan's questions, here is my current thinking on distributing Minix for the pc532. I would appreciate your comments. This is not final! Here's what I might distribute: min.Z, 103K, a minimal file system with all the files necessary for booting, plus a few commands minix.tar.Z, 252K, includes the OS image, nm output of the image so you can patch your disk info into it, all sources (not cdiff's) needed to recompile the image, libc.a, and a handful of other things After someone proves they have a legal copy of Minix by mailing me their official Minix boot disk with the Prentice-Hall copyright on it, I would return the disk and e-mail the above files to the individual. (Is e-mail secure enough for this? I would be sending plain-text copyrighted sources.) The above files could be divided into 7 pieces of less than 64K bytes each (is that the e-mail limit?) for e-mailing. I really would like to avoid copying floppies and mailing them by conventional mail. The files I have already mentioned do not include cc, as, ld, ar, ranlib, etc. However, these can be legally distributed by putting the binaries on the BU.EDU archive. Naturally, special arrangements will have to be made for those without access to the archive. The files I have now mentioned, in addition to a standard Minix distribution, are sufficient to build the whole works. The particular Minix version should not be too critical. I will post critical gotcha's, like how to fix fsck so it will not cream your disk. The latest Minix version is 1.5.10. It is supposed to be stable for quite some time because Dr. Tanenbaum has other things to do. However, it will not be distributed from Prentice-Hall until fall. Prentice-Hall currently is selling 1.3 (plus older versions -- beware). You can get 1.5.10 by buying some version from Prentice-Hall, getting an upgrade kit from an archive, and applying a million diff's. Or, better, get it from a friend who already did all that. I am aware that getting the latest Minix version is a real pain. However, it is a problem I do not intend to solve. I think you will do just fine with Minix-pc532, even if you do not have the latest version. The important thing is that you have the right sources for the OS image, which you get from the above procedure. My port is based on 1.3. I intend to upgrade it to 1.5.10, but I don't know when. I might have a look at it this weekend. So, what do you think? Bruce Culbertson
cagney@chook.ua.oz (Andrew Cagney - aka Noid) (06/09/90)
Sorry about being a bit `minix technical' From article <9006081924.AA02248@hplwbc.hpl.hp.com>, by culberts@hplwbc.hpl.hp.com (Bruce Culbertson): > The latest Minix version is 1.5.10. It is supposed to be stable > for quite some time because Dr. Tanenbaum has other things to > do. To the end user, this version is a significant improvement over 1.3 - many more commands (115 -> 160) - Much improved libc & include. - Additional software including UUCP is available for ftp/fetchfile. It has also seen a re-merge of the st & ibm versions. They are now identical except for the kernel proper. In addition, there is a 386 32bit version arround. (Again only the kernel is changed) > beware). You can get 1.5.10 by buying some version from Prentice-Hall, > getting an upgrade kit from an archive, and applying a million diff's. Fair comment. The process is very time consuming. In addition, the the less UNIX/C experience you have the slower it will be. > My port is based on 1.3. I intend to upgrade it to 1.5.10, but I > don't know when. I might have a look at it this weekend. Two key area's are effected 1. None of the existing 1.5 versions (ibm 386 atari mac amiga) understand single pages. The 386 32bit version for instance has taken a simple minded approach and allocates contigious memory to a process. The MM process would again need modifying to support page level allocation. 2. The Kernel, in places, was given a rewrite. This this includes how the kernel initially interacts with the MM & FS So to guess :-) I expect most of the work to be in the MM & Kernel. I'd also expect there to be a significant amount of work required. Most of it being the booring job of rearanging things so they interact correctly with the 1.5 MM & FS. Not a job for one weekend :-(. > latest version. The important thing is that you have the right > sources for the OS image, which you get from the above procedure. I don't know about the legalities of your distribution method but your justification is spot on. The last thing someone wants to be doing is spending weeks developing the cross developement environment, compiling MINIX, fixing compiler bugs... Having the patches available for ftp may also be useful. This would allow others (eg us here) with older NS32 versions of minix to catch up. Andrew Cagney
ian@sibyl.eleceng.ua.oz.au (06/09/90)
Jordan K. Hubbard writes: > I, like many of the folks on this list (I assume), am interested in > Bruce's minix port to the pc532 .... Has anyone thought about porting amoeba?
news@daver.bungi.com (06/09/90)
> > In response to Jordan's questions, here is my current thinking on > distributing Minix for the pc532. I would appreciate your comments. > This is not final! > > Here's what I might distribute: > > min.Z, 103K, a minimal file system with all the files > necessary for booting, plus a few commands > > minix.tar.Z, 252K, includes the OS image, nm output of > the image so you can patch your disk info into it, all > sources (not cdiff's) needed to recompile the image, > libc.a, and a handful of other things > > After someone proves they have a legal copy of Minix by mailing > me their official Minix boot disk with the Prentice-Hall copyright > on it, I would return the disk and e-mail the above files to the > individual. (Is e-mail secure enough for this? I would be sending [ stuff deleted] Possible alternative, assuming AST is agreeable. AST seems to have no problem with posting binaries and cdiffs to comp.os.minix. So what if we ask AST's permission to (1) post the above (except using cdiff's for source) to comp.os.minix and (2) permission to use comp.os.minix in the future as the "official" channel for PC532-Minix communications. (I am assuming these postings would in general be cross-linked to comp.sys.nsc.32k). Benefits: 1) Should eliminate immediate concerns of distributing copyrighted sources 2) Keeps AST informed/involved at all times (thinking of futures here) 3) Should attract additional interest in the pc532 concept/effort 4) Automaticly provides archive sites in North America, Europe, and Australlia (??) 5) etc.... Best regards, johnc PS. Some time back I sent an official Minix disk to Dave Rand as proof of purchase/license. --
cruff@ncar.ucar.edu (Craig Ruff) (06/09/90)
In article <9006081924.AA02248@hplwbc.hpl.hp.com> culberts@hplwbc.hpl.hp.com (Bruce Culbertson) writes: > minix.tar.Z, 252K, includes the OS image, nm output of > the image so you can patch your disk info into it, all > sources (not cdiff's) needed to recompile the image, > libc.a, and a handful of other things I think this will be the more useful of the two. >After someone proves they have a legal copy of Minix by mailing >me their official Minix boot disk with the Prentice-Hall copyright Is this really necessary? I'm considering getting the 1/2" tape version. Do you want me to mail the tape? Would it be sufficient to e-mail part of a program included in the distribution that hasn't been published? Perhaps the part should be chosen on a case by case basis. Well, that may be too much trouble. Of course, the Mac version may be available by then and I'll just get that instead. (Then mailing a floppy isn't much of a problem.) >on it, I would return the disk and e-mail the above files to the individual. E-mail is fine for me. Are the pieces included in minix.tar.Z really useful by themselves? For example, if it was just the kernel, mm, fs and 32k-specific parts of libc, it might make sense to place it into the archive too. A person would still need the rest of Minix to get things running. -- Craig Ruff NCAR cruff@ncar.ucar.edu (303) 497-1211 P.O. Box 3000 Boulder, CO 80307
Steven.D.Ligett@mac.dartmouth.edu (06/10/90)
--- You wrote:
...I might have a look at it this weekend.
^^^^^^^
So, what do you think?
--- end of quoted material ---
I think that comment recalls your earlier comment about how long it might take
to do the port, and recalls my comments on how quick I'd have parts kits to
sell!
Good luck!Mark-Geisert@l66a.ladc.bull.com (Mark Geisert) (06/10/90)
In a recent posting, Bruce Culbertson wrote: > After someone proves they have a legal copy of Minix by mailing > me their official Minix boot disk with the Prentice-Hall copyright > on it, I would return the disk and e-mail the above files to the > individual. (Is e-mail secure enough for this? I would be sending > plain-text copyrighted sources.) The above files could be divided > into 7 pieces of less than 64K bytes each (is that the e-mail > limit?) for e-mailing. I really would like to avoid copying floppies > and mailing them by conventional mail. To verify ownership of a legal copy of Minix, how about.. (1) prospective user snail-mails a photocopy of his Minix boot disk (with the P-H copyright on it) and his own e-mail address to Bruce (2) Bruce challenges prospective user by e-mail with a 'simple' question that only the holder of the Minix diskettes can answer (3) prospective user e-mails the answer to the question (or begs for another question?) (4) Bruce e-mails Minix materials. Re e-mail security, there is none :-). Perhaps you could leave out something(s) that are essential from your materials.. something that can be supplied by the holder of Minix diskettes. Just some ideas...... ..mark (Mark-Geisert@LADC.Bull.COM or ..!uunet!ladcgw!Mark-Geisert)
dlr@daver.bungi.com (Dave Rand) (06/10/90)
[In the message entitled "Re: Re: requested clarification of minix issues" on Jun 9, 11:53, Mark Geisert writes:]
> To verify ownership of a legal copy of Minix, how about..
The only way to "prove" ownership is to mail the _original_ disk, with
the P/H copyright notice on it. This is the only proof that is acceptable,
to me at least. No photocopies, "challanges", or anything else will suffice.
Bruce, George and I went over this issue a few months ago. Same goes for
tapes - send the original tape.
As a reminder: Andrew S. Tanenbaum wrote MINIX. He does not, however,
hold the copyright. Prentice-Hall holds the copyright. Even if AST tells
us that we can mass-duplicate it, and give it away - we still must confirm
this with Prentice-Hall's lawyers. We all know what they will say.
I will be happy to handle this, if no one else wants to.
--
Dave Rand
{pyramid|mips|sun|vsi1}!daver!dlr Internet: dlr@daver.bungi.comjonathan@comp.vuw.ac.nz (06/10/90)
[In the message entitled "Re: Re: requested clarification of minix issues"
on Jun 9, 11:53, Mark Geisert writes:]
> To verify ownership of a legal copy of Minix, how about..
The only way to "prove" ownership is to mail the _original_ disk, with
the P/H copyright notice on it. This is the only proof that is acceptable,
to me at least. No photocopies, "challanges", or anything else will suffice
What about if one's _original_ disk, with the P/H copyright
notice on it, was stolen?
This really happened here; does not having the original disk
leave me absolutely out in the cold?
How about constructing a patch kit containing cdifs for the NS32k
the minix kernel, mm, and fs, with new files as appropriate?
I'd be happy with diffs from *any* PC-minix version. And I do mean *any*:
1.1 (the original floppy or tape from PH)
1.2 (the second floppy version from PH)
1.3[a-f] (the pre-release kits or the third floppy)
1.4.x (never released by PH)
1.5.x (not yet released by PH)
for any x. (The only original PH floppies I had were 1.1)
I still have all these versions, or can at least re-generate
them from patch kits. I'm sure I'm not the only one.
Similar patch kits have been posted to Usenet by ast@cs.vu.nl; others
are available for anonymous FTP. I think this solution would
be far and away the easiest for those of us who *do* follow
comp.os.minix.
--Jonathan Stonedlr@daver.bungi.com (Dave Rand) (06/10/90)
[In the message entitled "Re: Re: Re: requested clarification of minix issues" on Jun 10, 13:07, jonathan@comp.vuw.ac.nz writes:] > [In the message entitled "Re: Re: requested clarification of minix issues" > on Jun 9, 11:53, Mark Geisert writes:] > > To verify ownership of a legal copy of Minix, how about.. > > The only way to "prove" ownership is to mail the _original_ disk, with > the P/H copyright notice on it. This is the only proof that is acceptable, > to me at least. No photocopies, "challanges", or anything else will suffice > > What about if one's _original_ disk, with the P/H copyright > notice on it, was stolen? > This really happened here; does not having the original disk > leave me absolutely out in the cold? A case such as yours can be handled on an individual basis. Still, to avoid copyright infringment, we must have some assurance that appropriate royalties have been paid. A receipt would suffice, or other written assurance. Besides, the MINIX disks are _very_ cheap; another copy can be purchased. I have done this on more than one occasion. Context diffs are an option for those that have a current working system, but to provide an easy "first start" for the people just starting, the current solution is the best. As we make changes to the system, context diffs make sense, and should be used. Bruce's modifications go beyond a two-line change though :-) Can we end this discussion now? On other topics, there has been a major upheaval at National Semiconductor. I do not know what the long term implication for the pc532 project is. It has delayed (again) some announcements from George and I... I'll keep you up to date. -- Dave Rand {pyramid|mips|sun|vsi1}!daver!dlr Internet: dlr@daver.bungi.com
culberts@hplwbc.hpl.hp.com (Bruce Culbertson) (06/11/90)
> --- You wrote: > ...I might have a look at it this weekend. > ^^^^^^^ > So, what do you think? > --- end of quoted material --- > I think that comment recalls your earlier comment about how long it might take > to do the port, and recalls my comments on how quick I'd have parts kits to > sell! I meant literally, "have a look". Just to get an idea of what the upgrade will involve. Actually, I did have a look and I do not think it will be too bad. But certainly more than a weekend. And I'm an optimist. Do you work at Kiewit? Did you work there in the early 1980's? Your name rings a bell. Good luck with your parts kits. Bruce
Steven.D.Ligett@mac.dartmouth.edu (06/12/90)
--- You wrote: Do you work at Kiewit? Did you work there in the early 1980's? Your name rings a bell. --- end of quoted material --- Yes, I think you mentioned in a note some months ago that you went to Thayer?? (I've been here for 14 years...)
news@daver.bungi.com (06/13/90)
Bruce: I'd like to put my 2c in on this discussion: > In response to Jordan's questions, here is my current thinking on > distributing Minix for the pc532. I would appreciate your comments. ... > After someone proves they have a legal copy of Minix by mailing > me their official Minix boot disk with the Prentice-Hall copyright > on it, I would return the disk and e-mail the above files to the > individual. (Is e-mail secure enough for this? I would be sending OK, seems silly ... > plain-text copyrighted sources.) The above files could be divided > into 7 pieces of less than 64K bytes each (is that the e-mail > limit?) for e-mailing. I really would like to avoid copying floppies > and mailing them by conventional mail. Then we have to download these files, then how do they get onto the scsi disk? OK. > The files I have already mentioned do not include cc, as, ld, ar, > ranlib, etc. However, these can be legally distributed by putting > the binaries on the BU.EDU archive. Naturally, special arrangements > will have to be made for those without access to the archive. Please remember to do this, as I am in that group. Now I understand we are using gcc and your assembler/linker/archiver? > The files I have now mentioned, in addition to a standard Minix > distribution, are sufficient to build the whole works. The > particular Minix version should not be too critical. I will I have 1.2. > I am aware that getting the latest Minix version is a real pain. > However, it is a problem I do not intend to solve. I think you I don't expect anyone can solve that problem except Prentice-Hall, and it appears that they are aware of it. All sounds very good. Minix 1.3 for PC is available at Walden Books for I think $79.