• Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.

In search of the A/ROSE development disks

eharmon

Well-known member
I've been futzing with A/ROSE and while I've found some of the development utilities on the general Apple development CDs, a bunch are missing.

Documentation references content "included on the Developer Essentials disc" and "included on the A/ROSE distribution disks", but I've never seen any archives of those. (https://vintageapple.org/develop/pdf/develop-04_9010_October_1990.pdf -- page 424).

It doesn't help "A/ROSE" is basically un-Googlable these days.

Has anyone seen a copy of these? Or happen to know a development CD that includes the full content?

Edit: Given the PDF, it seems like it comes from the October 1990 (volume 5?) version of the developer CD, a hodge-podge from 1990-91. The 1992 editions seem to have some but not all of the A/ROSE development components.
 
Last edited:

Andy

Well-known member
I don't have the software, but there are two different versions of the developer guide floating around on Bitsavers.
This is from Feb 1989 before the renamed MR-DOS to A/ROSE http://bitsavers.org/pdf/apple/mac/developer/MCP/MCP_Developers_Guide_198902.pdf
And this one is from November 89 http://bitsavers.org/pdf/apple/nubu...ocessor_Platform_Developers_Guide_Nov1989.pdf

Both pdfs contain a list of the files that were included on the distribution disks, So you can see what files you have and what's missing.

You might also try searching for "Macintosh Coprocessor Platform" as well as "MR-DOS." Though i doubt the latter with be any use.

I hope you're able to find something, A/ROSE is really fascinating. If i come across anything else interesting i'll post it here.
 

Melkhior

Well-known member
it seems like it comes from the October 1990 (volume 5?) version of the developer CD
That would be "night of the living disc" apparently. Can't find it online though :-(

EDIT: just realized it's not listed in the archive.org page, but is actually there! oups :)

EDIT2: A/ROSE 1.2.1 was available from https://staticky.com/dl/ftp.apple.com/developer/Tool_Chest/Devices_-_Hardware/, but that seems gone ?!? darn... Probably just the runtime stuff, but still... I'll put them here just in case.
 

Attachments

  • A_ROSETM_1.2.1.sit.hqx
    628.4 KB · Views: 0
  • A_ROSETM_1.2.1.txt
    2.4 KB · Views: 1
Last edited:

olePigeon

Well-known member
I feel like computers were advancing in speed too fast for A/ROSE to have ever really taken a hold, quickly becoming pointless. Well, that, and also because it was really expensive. However, I think it'd be nifty to utilize A/ROSE for modern, CPU-intensive stuff like USB that have so far eluded development on NuBus Macs. A multi-port NuBus ethernet card for running a vintage file server would be cool, too.

I mean, yeah, you could just offload stuff to an FPGA or something, but I think it would be really awesome to actually utilize A/ROSE for its intended purpose, even if it's also implemented in FPGA.
 

Melkhior

Well-known member
(..) modern, CPU-intensive stuff like USB that have so far eluded development on NuBus Macs (...)I mean, yeah, you could just offload stuff to an FPGA or something (...)
You could and 'someone' did :) Actually the problem for USB (and A/ROSE w/o its SDK and Ethernet using a modern chipset and, well, every other possible devices) is SW, not HW... NuBusFPGA has a USB connector and can have a full OHCI-compliant controller with no problem, same as in the SBusFPGA (and my TODO list include trying it with NetBSD/mac68k has NetBSD does have a proper stack and it works in NetBSD/sparc... but it's a long TODO list). But there's no software for it, and likely never will be in MacOS 8/System 7.

It _might_ have been possible to have enough USB support to get USB working in a NuBus PowerMac running MacOS 9 (registering an OHCI controller as could be done on PCI PowerMacs, creating just some NuBus-oriented glue), but the relevant SDK was per-request and not publicly available. I never found a trace of it on the Internet.

I feel like computers were advancing in speed too fast for A/ROSE to have ever really taken a hold, quickly becoming pointless
The default implementation of the MCP has a 10 MHz 68000 and 512 KiB or RAM. The Macintosh Plus was still being sold when the MCP was introduced! Apple though they could sell peripherals with a faster CPU than their entry-level system, including half the RAM - effectively a 'better' machine than an Atari 520ST or an Amiga 500... And Mac 512 were still being used... It was a stupidly expensive platform. Perhaps it could have made sense for some high-end stuff in data acquisition/processing, but for an Ethernet device? The NE2000 already existed at that time if you wanted cheap, and the Am7990 ('Lance Ethernet', used by Sun among many, many others) was long available for higher-end implementation. Neither needed a full CPU and 512 KiB of RAM. I my opinion, A/ROSE was pointless from day 1, if not before. One of Apple's many inane decisions of the era.

If you ignore this business aspect, it was an awesome design and it created the first true 'smart NiC', decades before they became popular and (perhaps) cost-effective. And as 68000 are available as soft-core from various projects, I did wonder about recreating a virtual MCP with the NuBusFPGA... but without all the proper original software, it's pointless :-(
 

Melkhior

Well-known member
Would it be possible to implement NetBooting via a combination of A/ROSE and ROM?
I don't think A/ROSE would even be necessary - there's a thread on netbooting there that suggests it's possible on some Macs.

NuBus devices (and PDS) can be used to boot from 'disks' (or anything they will present to MacOS as such), so a device could create a "virtual disk" by any means necessary and let the machine boot from that, presumably including remote filesystems of one form or another. A/ROSE might have made it easier to implement the client-side of such an implementation, but again might not have been really necessary - the real-time aspect is overkill. A device could also load a remote image into some local memory, and then expose such memory as a bootable drive - perhaps easier to implement but make the bootable disk non-modifiable between boots. Again, no need for real-time, the 'R' from A/ROSE...

BTW out of curiosity, why would you want netbooting in 2023? Centralizing booting made sense back in the day when storage was super expensive (hence Sun inventing NFS and enabling workstations to net-boot, net-root & net-swap over NFS), but by now it's probably easier to just stick a micro-sd card in the machine. Another thing on my TODO list for NuBusFPGA, fix the micro-sd support...
 

cheesestraws

Well-known member
It's mostly, tbh, that Apple were just never very good at coprocessors. It was always an afterthought, difficult to code for, and not really very useful, especially as main CPUs got faster.
 

eharmon

Well-known member
By combining the A/ROSE 1.2.1 development directories from Developer CDs, a hodge podge of MPW updates, and the examples from Dev CD V, I think I have a mostly working development environment and I've (sort of) managed to load some code onto my card. It's nothing if not unstable though. And loves to take out the host system.

The only thing I've not found by those sources combined is a later version of NuBug. The card LOVES to crash when you do anything funky so if anyone has any versions besides the very earliest, let me know!

Once I get a clean set of code downloaded to the card I'm gonna write up a guide for anyone else who wants to futz around.
 
Top