mpierce@ewu.UUCP (Mathew Pierce) (04/04/91)
I hear that the A3000UX is ABI compatible. I heard that this gives it binary compatibility with certain applications which also conform to this standard, is this correct? If so, is autocad ABI compatible, and if so does that mean it will run on the A3000UX? Could someone post a list of ABI applications that the A3000UX can run, and anything else that it can run? Matt Pierce
jesup@cbmvax.commodore.com (Randell Jesup) (04/05/91)
In article <1522@ewu.UUCP> mpierce@ewu.UUCP (Mathew Pierce) writes: >I hear that the A3000UX is ABI compatible. I heard that this gives >it binary compatibility with certain applications which also >conform to this standard, is this correct? I believe 680x0 ABI programs should run on it, though I'm an AmigaDos guy and don't know for sure. -- Randell Jesup, Keeper of AmigaDos, Commodore Engineering. {uunet|rutgers}!cbmvax!jesup, jesup@cbmvax.commodore.com BIX: rjesup Disclaimer: Nothing I say is anything other than my personal opinion. Thus spake the Master Ninjei: "To program a million-line operating system is easy, to change a man's temperament is more difficult." (From "The Zen of Programming") ;-)
khh@root.co.uk (Keith Holder) (04/08/91)
In <1522@ewu.UUCP> mpierce@ewu.UUCP (Mathew Pierce) writes: >I hear that the A3000UX is ABI compatible. I heard that this gives >it binary compatibility with certain applications which also >conform to this standard, is this correct? If so, is autocad ABI We have an Amiga in the office running Commodore's 1.1 version of System VR4 and I would say that it is not ABI compliant. We also have various other 68030 machines running V.4 from Motorola and binaries taken from these machines and run on the Amiga box result in core dumps. Looking at various header files on the Amiga and comparing them against the Motorola 68k ABI, I think that they have quite a way to go before they are ABI compatible. keith holder. Obviously, these are my opinions and do not reflect the opinions of my employers who just so happen to produce a version of V.4 for the Motorola 68k. -- -- Keith Holder, Systems Software Consultant, UniSoft Ltd. <khh@root.co.uk> G1ITH Fax: (071) 729 3273 Phone: +44 71 729 3773
david@twg.com (David S. Herron) (04/09/91)
In article <1522@ewu.UUCP> mpierce@ewu.UUCP (Mathew Pierce) writes: >I hear that the A3000UX is ABI compatible. I heard that this gives >it binary compatibility with certain applications which also >conform to this standard, is this correct? If so, is autocad ABI >compatible, and if so does that mean it will run on the A3000UX? ABI can only specify compability among Unix on processors of a certain family. In this case only among 680x0 based Unix boxen. I don't know of many in that family which are ABI, but there probably are a few. You will not (obviously?) be able to take 80386 ABI binaries over to an Amiga Unix system and do anything useful with them... A possible reason for the rumor floating about last fall of discussions 'tween Sun & C= is C= wanting Amiga Unix to be able to run SunOS binaries. *THAT* would be *useful* ... ABI specifies the `numbers' for the system calls & what system calls are there & what arguments they take & such. I suppose it also specifies things about shared libraries and who knows what else. David -- <- David Herron, an MMDF & WIN/MHS guy, <david@twg.com> <- Formerly: David Herron -- NonResident E-Mail Hack <david@ms.uky.edu> <- <- "MS-DOS? Where we're going we don't need MS-DOS." --Back To The Future
es1@cunixb.cc.columbia.edu (Ethan Solomita) (04/09/91)
In article <8823@gollum.twg.com> david@twg.com (David S. Herron) writes: > >I don't know of many in that family which are ABI, but there probably >are a few. You will not (obviously?) be able to take 80386 ABI >binaries over to an Amiga Unix system and do anything useful >with them... > BUT, under SVR4 all that is necessary to port the program, from Sun for example, is a mere recompilation. Unless the program does something very unusual, X makes for a useful standard. -- Ethan Q: How many Comp Sci majors does it take to change a lightbulb A: None. It's a hardware problem.
tbissett@nstar.rn.com (Travis Bissett) (04/09/91)
david@twg.com (David S. Herron) writes: > In article <1522@ewu.UUCP> mpierce@ewu.UUCP (Mathew Pierce) writes: > >I hear that the A3000UX is ABI compatible. I heard that this gives > >it binary compatibility with certain applications which also > >conform to this standard, is this correct? If so, is autocad ABI > >compatible, and if so does that mean it will run on the A3000UX? > > ABI can only specify compability among Unix on processors of > a certain family. In this case only among 680x0 based Unix boxen. > To chip in with my $.02 worth . . . I asked the Autodesk rep at the National Design Show in Chicago today if AutoCAD will support Unix SVr4 with the MC68K ABI -- I specifically mentioned the HP/Apollo line, older Suns, and (of course) the Amiga 300UX. His response was no -- Autodesk has no plans to support unix workstations other than 386 Unix and SPARC unix, with both 386 and Sun IPC platforms being so "affordable." Well, too bad for them! Travis P.S. read A3000UX -- dumb keys! ;-) -- Travis Bissett NSTAR conferencing site 219-289-0287 internet: tbissett@nstar.rn.com 1300 newsgroups - 8 inbound lines uucp: ..!uunet!nstar.rn.com!tbissett 99 file areas - 4300 megabytes --- backbone news & mail feeds available - contact larry@nstar.rn.com ---
es1@cunixb.cc.columbia.edu (Ethan Solomita) (04/09/91)
In article <2671@root44.co.uk> khh@root.co.uk (Keith Holder) writes: >In <1522@ewu.UUCP> mpierce@ewu.UUCP (Mathew Pierce) writes: > >>I hear that the A3000UX is ABI compatible. I heard that this gives >>it binary compatibility with certain applications which also >>conform to this standard, is this correct? If so, is autocad ABI >We have an Amiga in the office running Commodore's 1.1 version of >System VR4 and I would say that it is not ABI compliant. We also >have various other 68030 machines running V.4 from Motorola and >binaries taken from these machines and run on the Amiga box >result in core dumps. Looking at various header files on the >Amiga and comparing them against the Motorola 68k ABI, I think >that they have quite a way to go before they are ABI compatible. There is some other problem because AMIX is ABI compliant. Also, I don't seem to remember hearing about a version 1.1, just 1.0, although I could be wrong about that. > keith holder. > >Obviously, these are my opinions and do not reflect the opinions of my >employers who just so happen to produce a version of V.4 for the Motorola 68k. >-- >-- >Keith Holder, Systems Software Consultant, UniSoft Ltd. ><khh@root.co.uk> G1ITH Fax: (071) 729 3273 >Phone: +44 71 729 3773 -- Ethan Q: How many Comp Sci majors does it take to change a lightbulb A: None. It's a hardware problem.
dltaylor@cns.SanDiego.NCR.COM (Dan Taylor) (04/10/91)
In <2671@root44.co.uk> khh@root.co.uk (Keith Holder) writes: >We have an Amiga in the office running Commodore's 1.1 version of System VR4 >and I would say that it is not ABI compliant. We also have various other >68030 machines running V.4 from Motorola and binaries taken from these machines And EXACTLY which System V.4 ABI CERTIFIED operating systems are you running, and on which platforms? Since I have been unable to get information on ANY OTHER such systems available commercially in the U.S. (and I have been to more trade shows, and in more reps offices than I care to count), I wonder where you got them, and when they were CERTIFIED. >employers who just so happen to produce a version of V.4 for the Motorola 68k. And YOURS is CERTIFIED ABI? Your companies products are consistantly incompatible, even within an AT&T release version. Thank (fill in blank) that C= had more sense than to use YOU. More often than not, I've cursed Unisoft for putting out what I consider ABSOLUTE JUNK releases of V.3. >Keith Holder, Systems Software Consultant, UniSoft Ltd. It is possible that C= slipped up, and that the Amiga distribution of UNIX is not ABI compliant, or that 2.0 will be cleaned up as a planned product. However, unless you can document your statement, SHUT the #$#^%$# UP! Dan Taylor /* My opinions, not NCR's. */
ajc@root.co.uk (Tony Cleverley) (04/10/91)
In <1991Apr9.063955.12000@cunixf.cc.columbia.edu> es1@cunixb.cc.columbia.edu (Ethan Solomita) writes: >In article <2671@root44.co.uk> khh@root.co.uk (Keith Holder) writes: >>We have an Amiga in the office running Commodore's 1.1 version of >>System VR4 and I would say that it is not ABI compliant. We also >>have various other 68030 machines running V.4 from Motorola and >>binaries taken from these machines and run on the Amiga box >>result in core dumps. Looking at various header files on the >>Amiga and comparing them against the Motorola 68k ABI, I think >>that they have quite a way to go before they are ABI compatible. > There is some other problem because AMIX is ABI >compliant. I think you are wrong as far as I can see the AMIGA 3000 UX is not compliant, for example: The current ABI defines the structure sigaction (see sys/signal.h) to be: struct sigaction { void (*sa_handler)(); sigset_t sa_mask; int sa_flags; } yet the AMIGA defines it as: struct sigaction { int sa_flags; void (*sa_handler)(); sigset_t sa_mask; int sa_resv[2]; } Also the online manual page for sigaction on the AMIGA states: The sigaction structure includes the following members: void (*sa_handler)(); sigset_t sa_mask; int sa_flags; There is no mention of int sa_resv[2]; >Also, I don't seem to remember hearing about a version >1.1, just 1.0, although I could be wrong about that. On the AMIGA we have here, uname -a gives UNIX_Sytem_V localhost 4.0 1.1 Amiga m68020 This could be wrong though as the sytem here has a m68030 running at 25MHz. Tony Cleverley Obviously, these are my opinions and do not reflect the opinions of my employers who just so happen to produce a version of V.4 for the Motorola 68k. Tony Cleverley, Systems Engineer, UniSoft Ltd. <ajc@root.co.uk> G1ITH Fax: (071) 729 3273 Phone: +44 71 729 3773
dfields@radium.urbana.mcd.mot.com (David Fields) (04/11/91)
In article <8823@gollum.twg.com>, david@twg.com (David S. Herron) writes: |>ABI specifies the `numbers' for the system calls & what system |>calls are there & what arguments they take & such. I suppose it |>also specifies things about shared libraries and who knows what else. Well not really. The ABI is split into a generic section and an architecture specific section. It specifies among other things, application packaging and installation, call return sequences, the memory map for a process and header file information neccessary for compatability between different vendors shared libraries. An ABI application must be dynamically linked with a subset of libc reffered to as libsys. Libsys contains all of the system calls available to ABI compliant program. In other words, an ABI compliant program cannot contain a direct system call which means neither system call numbers nor even mechanisms need be the same on ABI compliant systems. The ABI's are a great idea but do require alot of work by the respective companies to find and fix compatability problems. This can be seen by the amount of work being but into the 88k ABI effort via 88open and it's members. Dave Fields // Motorola Computer Group // dfields@urbana.mcd.mot.com
khh@root.co.uk (Keith Holder) (04/11/91)
In <889@cns.SanDiego.NCR.COM> dltaylor@cns.SanDiego.NCR.COM (Dan Taylor) writes: >And EXACTLY which System V.4 ABI CERTIFIED operating systems are you >running, and on which platforms? Since I have been unable to get information >on ANY OTHER such systems available commercially in the U.S. (and I have been >to more trade shows, and in more reps offices than I care to count), I >wonder where you got them, and when they were CERTIFIED. UniSoft not only wrote the AT&T System VR4 68k/88k product, they also developed the gABI test suite for AT&T, so maybe we have much more of an idea of what is ABI compliant and what is not. Also you don't need to be a genius to compare the header files on the Amiga box and what is specified in the Motorola 68k ABI and spot inconsistencies in data structures. Plus, I am not aware of an ABI certification procedure that is in place for any VR4 release on any processor architecture. >>employers who just so happen to produce a version of V.4 for the Motorola 68k. >And YOURS is CERTIFIED ABI? Your companies products are consistantly >incompatible, even within an AT&T release version. Thank (fill in blank) I don't remember saying that our product was certified as ABI compliant, however, I do remember saying that the Amiga UNIX , IMHO, was not ABI compliant. I could also prove it by indicating which header files on the Amiga system that we have are incompatible with the ABI specification. I also know that we have followed the 68k ABI as much as humanly possible, even when the goal-posts have been moved. >that C= had more sense than to use YOU. More often than not, I've cursed >Unisoft for putting out what I consider ABSOLUTE JUNK releases of V.3. Oh yeah, have you certified evidence that they were junk? And by the way since we are getting personal here, I have never considered NCR's release of UNIX that wonderful either. >It is possible that C= slipped up, and that the Amiga distribution of UNIX >is not ABI compliant, or that 2.0 will be cleaned up as a planned product. Really? No shit, Sherlock! >However, unless you can document your statement, SHUT the #$#^%$# UP! Why should I? I was only offering my humble opinion to a question on whether Amiga Unix was ABI compatible. All you have done is air your grievances about past UniSoft products. Just because you have an axe to grind with UniSoft, ( did they turn down a job application? ) you seem to think that their employees have no right to express an opinion on the net. So, until you can provide documentary evidence that Amiga Unix is ABI compatible maybe you should shut the f**k up! B-). -- -- Keith Holder, Systems Software Consultant, UniSoft Ltd. <khh@root.co.uk> G1ITH Fax: (071) 729 3273 Phone: +44 71 729 3773
swansonc@acc.stolaf.edu (Chris Swanson) (04/13/91)
>>>>> On 11 Apr 91 11:08:39 GMT, >>>>> in message <2675@root44.co.uk>, >>>>> khh@root.co.uk (Keith Holder) wrote: khh> In <889@cns.SanDiego.NCR.COM> dltaylor@cns.SanDiego.NCR.COM (Dan khh> Taylor) writes: >And EXACTLY which System V.4 ABI CERTIFIED operating systems are you >running, and on which platforms? Since I have been unable to get >information on ANY OTHER such systems available commercially in the >U.S. (and I have been to more trade shows, and in more reps offices >than I care to count), I wonder where you got them, and when they >were CERTIFIED. khh> UniSoft not only wrote the AT&T System VR4 68k/88k product, khh> they also developed the gABI test suite for AT&T, so maybe we khh> have much more of an idea of what is ABI compliant and what is khh> not. Also you don't need to be a genius to compare the khh> header files on the Amiga box and what is specified in the khh> Motorola 68k ABI and spot inconsistencies in data structures. khh> Plus, I am not aware of an ABI certification procedure that is in khh> place for any VR4 release on any processor architecture. [deleted drivel] khh> Really? No shit, Sherlock! >However, unless you can document your statement, SHUT the #$#^%$# UP! khh> Why should I? I was only offering my humble opinion to a khh> question on whether Amiga Unix was ABI compatible. All you have khh> done is air your grievances about past UniSoft products. Just khh> because you have an axe to grind with UniSoft, ( did they turn khh> down a job application? ) you seem to think that their employees khh> have no right to express an opinion on the net. So, until you can khh> provide documentary evidence that Amiga Unix is ABI compatible khh> maybe you should shut the f**k up! B-). Do we really need the low S/N of this little trade war (NCR vs. UniSoft) here on c.u.a? Why don't you both put your axes away and shake hands (or at least go to seperate corners). This should really be continued via e-mail between you too if you feel more "discussion" etc. are necessary. -Chris -- Chris Swanson, Chem/CS/Pre-med Undergrad, St. Olaf College, Northfield,MN 55057 DDN: (CDS6) INTERNET: swansonc@acc.stolaf.edu UUCP: uunet!stolaf!swansonc AT&T: Work: (507)-645-4528 Home: (507)-663-6424 I would deny this reality, but that wouldn't pay the bills...
swansonc@acc.stolaf.edu (Chris Swanson) (04/13/91)
>>>>> On 11 Apr 91 11:08:39 GMT, >>>>> in message <2675@root44.co.uk>, >>>>> khh@root.co.uk (Keith Holder) wrote: khh> In <889@cns.SanDiego.NCR.COM> dltaylor@cns.SanDiego.NCR.COM (Dan khh> Taylor) writes: >And EXACTLY which System V.4 ABI CERTIFIED operating systems are you >running, and on which platforms? Since I have been unable to get >information on ANY OTHER such systems available commercially in the >U.S. (and I have been to more trade shows, and in more reps offices >than I care to count), I wonder where you got them, and when they >were CERTIFIED. khh> UniSoft not only wrote the AT&T System VR4 68k/88k product, khh> they also developed the gABI test suite for AT&T, so maybe we khh> have much more of an idea of what is ABI compliant and what is khh> not. Also you don't need to be a genius to compare the khh> header files on the Amiga box and what is specified in the khh> Motorola 68k ABI and spot inconsistencies in data structures. khh> Plus, I am not aware of an ABI certification procedure that is in khh> place for any VR4 release on any processor architecture. [deleted drivel] khh> Really? No shit, Sherlock! >However, unless you can document your statement, SHUT the #$#^%$# UP! khh> Why should I? I was only offering my humble opinion to a khh> question on whether Amiga Unix was ABI compatible. All you have khh> done is air your grievances about past UniSoft products. Just khh> because you have an axe to grind with UniSoft, ( did they turn khh> down a job application? ) you seem to think that their employees khh> have no right to express an opinion on the net. So, until you can khh> provide documentary evidence that Amiga Unix is ABI compatible khh> maybe you should shut the f**k up! B-). Do we really need the low S/N of this little trade war (NCR vs. UniSoft) here on c.u.a? Why don't you both put your axes away and shake hands (or at least go to seperate corners). This should really be continued via e-mail between you too if you feel more "discussion" etc. is necessary. -Chris -- Chris Swanson, Chem/CS/Pre-med Undergrad, St. Olaf College, Northfield,MN 55057 DDN: (CDS6) INTERNET: swansonc@acc.stolaf.edu UUCP: uunet!stolaf!swansonc AT&T: Work: (507)-645-4528 Home: (507)-663-6424 I would deny this reality, but that wouldn't pay the bills...
cy0q+@andrew.cmu.edu (Chad O. Yoshikawa) (04/13/91)
Keith, Maybe the two of you should calm down a little bit! Geez! Hold on, I'll get a spare pair of my boxing gloves and we'll make some wagers.... Anyway, the only information I can offer is that the magazine advertisement did say that the 3000/ux was abi compliant. Chad
dltaylor@cns.SanDiego.NCR.COM (Dan Taylor) (04/16/91)
First, an apology to Keith, and the net: I shouldn't have reacted so strongly to Keith's original posting. I apologize. In <2675@root44.co.uk> khh@root.co.uk (Keith Holder) writes: > UniSoft not only wrote the AT&T System VR4 68k/88k product, they also >developed the gABI test suite for AT&T, so maybe we have much more of an idea >of what is ABI compliant and what is not. What AT&T System VR4 68k product? > Also you don't need to be a genius to compare the header files on the >Amiga box and what is specified in the Motorola 68k ABI and spot >inconsistencies in data structures. Your original posting said "core dumps". If you run a non-compliant program from platform "A" on platform "B", and they don't use the same load addresses, trap vector to the system, or system call interface, it will probably core dump. So, unless you run a compliant program, and it core dumps, you have no complaint. You didn't say you did that. You COULD have even posted that it dumped because it "used the wrong vector", or whatever. The ABI was purposely kept loose, to permit non-compliant software to continue to run on its original platform, on as many platforms as possible. So "running programs from different systems" is meaningless, unless each of those programs is known to be compliant. >I could also prove it by indicating which header files on the Amiga system that >we have are incompatible with the ABI specification. Then why didn't you? > Why should I? I was only offering my humble opinion to a question on >whether Amiga Unix was ABI compatible. No, you offered an "authoritative statement", no qualifiers, no documentation. That is not constructive, for anyone, but sounds like "advertising", especially when you talk about Unisoft "providing a System V". I HATE advertising on the net. However, I was too strong. >( did they turn down a job application? ) It would never have entered my mind to apply. I have worked in UNIX (kernel and apps) for a decade, at least. But I've not worked in Europe in that capacity. If Unisoft has/had a Left Coast, USofA, office, I've not bothered to look for it. >So, until you can provide >documentary evidence that Amiga Unix is ABI compatible maybe you should shut >the f**k up! B-). >Keith Holder, Systems Software Consultant, UniSoft Ltd. I provided as much evidence that it was as you did against, in your original posting. You made an unsupported, at the time, statement that C='s UNIX was not ABI-compliant. Had you provided ANY documentation, then I would not have complained. One of the purposes of these groups is education. If you're going to complain about something, document the complaint. If a header is bad, list the offense. If a binary crashes, get out your favorite debugger, and say where. My real complaint with your posting was that you just "flamed" C=. That's why I "demanded" some backup documentation, or silence. Also, it may be that you have out-of-date code. Have you reported the problem to C=? If they have already fixed this, then there IS no problem. I would have to get back to several-year-old history files to find my exact complaints against Unisoft ports. I should not have brought up my prior dissatisfactions without documentation, any more than you should have posted without it. It may be that since I last used one, new, or more experienced people, are doing the job, and those problems no longer exist. I had nothing to do with NCR's UNIX in the past, and am a consultant here. That's why the disclaimer exists. Dan Taylor * My opinions, not NCR's. *
ford@amix.commodore.com (Mike "Ford" Ditto) (05/14/91)
Now that the flame fest has died down for a while, I'll try to explain the issue of ABI compatibility and Amiga Unix. Amiga Unix is advertised as being ABI compatible. That is, we "guarantee" that you can take an ABI compatible binary program and run it on an Amiga Unix system. We also provide software development tools that are intended to create ABI compatible binary programs, and we actively persue and resolve any complaints or claims of ABI incompatibility regarding any part of our system. We have not and will not intentionally cause Amiga Unix to be incompatible with the m68k ABI. At the present time, however, it is somewhat difficult to determine exactly what is the "official" ABI. For example, several changes have been made by Motorola since the "blue book" m68k ABI manual was published by AT&T/Unix Press. There are other issues and proposed changes still being resolved that will almost certainly change the ABI further. Because of this fact, we have not yet conducted an exhaustive study verifying compatibility between Amiga Unix and any particular revision of the m68k ABI. We intend to persue the ABI issue with Motorola, USL, and other parties interested in the m68k ABI. When the "final" m68k ABI exists, Amiga Unix will be compatible with it. Hopefully this will be in time for our version 2.0. es1@cunixb.cc.columbia.edu (Ethan Solomita) writes: > There is some other problem because AMIX is ABI compliant. ajc@root.co.uk (Tony Cleverley) writes: > I think you are wrong as far as I can see the AMIGA 3000 UX is not compliant, > for example: > > The current ABI defines the structure sigaction (see sys/signal.h) > to be: > > struct sigaction { > void (*sa_handler)(); > sigset_t sa_mask; > int sa_flags; > } > > yet the AMIGA defines it as: > > struct sigaction { > int sa_flags; > void (*sa_handler)(); > sigset_t sa_mask; > int sa_resv[2]; > } This particular header file in Amiga Unix comes directly from the group in Motorola which maintains the official ABI specification. It turns out that there was a mistake and that this structure is in fact incorrect. This is a significant incompatilibity and essentially means that Amiga Unix version 1.1 is not able to run most ABI programs. It, along with any other known ABI incompatibilities, will be fixed in a future version of Amiga Unix. -=] Ford [=- "The heart and the mind on (In Real Life: Mike Ditto) a parallel course, never ford@amix.commodore.com the two shall meet." uunet!cbmvax!ditto -- Indigo Girls ford@kenobi.commodore.com