A/UX distinctives...

[ChristTrekker]

No position independent code. Static libraries are given a fixed address at compile time and can't conflict with other shared libraries. You can only export functions, not data, hence there is a shared library for X11 but not for Xt.

OK, I imagine that's a significant limitation, but in a practical sense I have no idea what kind of libraries would be affected and which not. I guess I'll just be resigned to building everything on A/UX statically.

"X/Open Transport Interface" and "Transport Layer Interface"

Is similar to BSD sockets. Not really used any more.

Is that what replaced it?

How about the "date" command won't accept any date after 1999?

Well, yeah, that's pretty significant, but that's only one command. Anything date-oriented in the system still works fine with dates after 2000 (AFAIK). I wrote a primitive dateset utility to hack around this. I keep thinking that someday I'll port a better date command to A/UX...someday...

Commando

I must admit I have not looked into it, I imagine it's similar to MPW.

Seems to me that all you'd need is some kind of database that list the options, allowed values, and possible combinations thereof.

 

[Gorgonops]

Commando

I must admit I have not looked into it, I imagine it's similar to MPW.

Seems to me that all you'd need is some kind of database that list the options, allowed values, and possible combinations thereof.

Just as a useless aside, it might be "interesting" if someone with way too much time on their hands would look into dissecting the version of System 7 that A/UX runs in its compatibility box. (And of course as its main interface, unless you start in X11/terminal mode.) It seems to me that if you could define the interfaces which A/UX uses to pass information in and out of the virtual machine it should in principle hack the BasiliskII emulator to support the same message passing mechanism. (You'd of course have to write an accessory daemon to generate/interpret messages appropriately.) This would let you run the A/UX System and accessory programs on a more modern UNIX/Linux, discarding the archaic undercarriage of A/UX entirely. And it's the Macintosh System-y parts of A/UX that most people actually care about.

(And of course it'd also let you use modern hardware. BasiliskII with JIT on a modern PC is many times faster then the fastest Quadra ever made, and of course the UNIX backend will run at full native speed.)

The end result of this would basically be a cross-platform-able version of MAE. (You'd need different syntax databases for "commando" and whatnot for each UNIX you intend to support, of course) So far as that goes, I'm willing to bet that the Mac emulator included in MAE used very similar if not identical patches to the System/Mac Toolbox for passing messages, and by patching BasiliskII to support the A/UX 7.whatever System it would be a fairly short leap to getting the System 7.5.3 included with MAE 3.0 working. (Assuming you can extract it from the trials floating around, and get it running without a license key. My guess is the licensing system with MAE only affects whether the container emulator will run or not, but I dunno. And undoubtedly it'd still be "pirating" to run the MAE specific bits without an MAE license. But it's pirating for most people to run A/UX, so...)

And of course, if you really want to go nuts you could then patch the same message passing system into SheepShaver and use MacOS 9.0.4 as your UNIX frontend. (You'd of course have to write PowerPC versions of whatever A/UX-ish bits you wanted, or run the old 68k binaries after patching them so they don't freak out after being moved to your completely artificial environment.)

Anyway.

 

[porter]

This would let you run the A/UX System and accessory programs on a more modern UNIX/Linux, discarding the archaic undercarriage of A/UX entirely. And it's the Macintosh System-y parts of A/UX that most people actually care about.

There isn't really anything special that runs with the System 7 layer except a command line and a load of control panels that say don't change me here, change me in A/UX proper.

I've run A/UX "headless" on a number of machines IIcx, IIvx, Q700, I've never run any Mac applications on them. Mac X will run on standard MacOS.

 

[Gorgonops]

There isn't really anything special that runs with the System 7 layer except a command line and a load of control panels that say don't change me here, change me in A/UX proper.
I've run A/UX "headless" on a number of machines IIcx, IIvx, Q700, I've never run any Mac applications on them. Mac X will run on standard MacOS.

"Commando" is about the only "interesting" hybrid application I remember from A/UX, granted. (I gave away the hardware to run it about five years ago.) There's also the UNIX filesystem/MacOS integration, I guess, but that's covered by Basilisk already. I'm mostly curious exactly how Commando worked... it seems in principle it might of worked simply by opening a telnet port to localhost with no special message passing system at all, but I never dug that deeply into it.

If one were to take the approach of "recreating" A/UX by using BasiliskII or SheepShaver an "obviously" better way of handling X11 programs then running crufty old MacX would be to write an application that interacts with the emulator to allow native X11 windows to selectively "punch through" a full-screen "always on top" Mac desktop. Basically I picture something along the lines of running "Sheepshaver/UX" session as your Window Manager. You'd have a MacOS application which would act as a launchpad for X11 apps, and when one was launched a blank window would be created on the Mac desktop by your MacX replacement. Then said MacX replacement would inform the emulator to mask out the appropriate area allowing the X11 app to "peek through", and would route keyboard and mouse actions to the active UNIX window when it has the focus. The deep-in details of tiling with native Mac programs and handling the clipping for "non-square" window objects would undoubtedly get complicated, but in principle it could work.

Given the dead-ness of classic MacOS as an application platform such a project would be an incredible waste of time, no denying that. (Unless you're in the "I'm never giving up OS 9 until they pry it from my cold dead fingers" camp it's a lot of work just to run a decade-old version of Photoshop alongside Firefox 3.5 in a seamless window.) It's mostly just an argument that it *could* be doable without a "huge" development project. Minimalistic GPL or BSD-licensed window managers are a dime a dozen so there's plenty of sources of code to reference when patching window management functionality into Sheepshaver or BasiliskII. As for the MacOS application, the bulk of its job would be tracking blank windows, updating title bars, and keeping the emulator informed about what regions to apply the clipping mask to. How hard that would be to write I haven't the foggiest.

As to running A/UX headless I have no doubt you can. It's just not what gets people hyped about A/UX. There are plenty of better *NIXes to run headless these days unless you have a special love for genuine (and outdated) AT&T intellectual property. There's always SIMH's PDP-11 and VAX emulations if you want *real* hardcore ancient UNIX thrills.

 

[porter]

If one were to take the approach of "recreating" A/UX

No need, we have Mac OS X. Unix on the bottom, Mac on top, supports Macintosh file systems.

As to running A/UX headless I have no doubt you can.

There is no doubt about it! I've even run it on a IIvx headless. I do collect UNIXes, and have AIX on i386 & PPC, AUX, HPUX on PARISC, SunOS on 68k & Sparc, Solaris on Sparc & i386, IRIX, Linux and the various *BSDs on MIPS, i386, 68k & PPC, and of course base Darwin installs on i386 & PPC.

There's always SIMH's PDP-11 and VAX emulations if you want *real* hardcore ancient UNIX thrills.

Yes, done that with real Release 7 UNIX, 256k memory, 4 x10meg harddisks and a simulated tapedrive!

 

[ChristTrekker]

There is no doubt about it! I've even run it on a IIvx headless. I do collect UNIXes, and have AIX on i386 & PPC, AUX, HPUX on PARISC, SunOS on 68k & Sparc, Solaris on Sparc & i386, IRIX, Linux and the various *BSDs on MIPS, i386, 68k & PPC, and of course base Darwin installs on i386 & PPC.

You're even crazier than I am.

 

[Gorgonops]

There is no doubt about it! I've even run it on a IIvx headless. I do collect UNIXes, and have AIX on i386 & PPC, AUX, HPUX on PARISC, SunOS on 68k & Sparc, Solaris on Sparc & i386, IRIX, Linux and the various *BSDs on MIPS, i386, 68k & PPC, and of course base Darwin installs on i386 & PPC.

Once upon a time I had a pretty good collection myself, although I have to admit that beyond a certain point the novelty started wearing off. ("Hrm, why exactly am I wasting so much time on this when I could just slap Debian or Centos on a cheap PC brick and type 'apt-get' or 'yum' a few times and be done with it? Oh, yeah, the noble challenge of it. Yeah, that's it...") Now it pretty much has to be genuinely fun, puzzling, or work related to get me to spend time on it anymore. ;^b

Yes, done that with real Release 7 UNIX, 256k memory, 4 x10meg harddisks and a simulated tapedrive!

I had BSD 2.11 on a virtual PHP-11 telnet-able on the company network for a few months years ago. (Looked to be about the oldest thing out there with a decent tcp/ip stack.) The old salts used to get a kick out of it.