Classic AmigaOS On Modern Hardware - A Critical Analysis

[matthey]

Indeed. What hardware do you run? I'd be happy to tell you if its supported. Also HAMMERFS on either MorphOS or AmigaOS 4 would be awesome - it would bring ZFS like behaviours to Amigans. Imagine: snapshots, logical volumes, software RAID - the possibilities!

I don't currently have a Linux/BSD box setup but I have a spare Pentium 4 (with hyper- threading and 2GB of memory) box with Intel chipset and a new half Terrabyte SATA HD. It's nothing special but I figured I could try out Linux/BSD/AROS/Haiku etc. on it if I could find the time. I currently have my Pentium M Windows XP laptop networked to my Amiga using SMBFS. It gives me AmiDevCpp compiling, UAE, modern web browsing and a general file server. It's efficient and sufficient for my needs but it wouldn't hurt to have a backup computer. I have some Linux experience but I would like to avoid some of the time consuming Linux tediousness like hardcore shell usage, endless settings configurations and the need to compile everything. I would rather stay away from needless eye candy bloat and degenerative GUIs that require many clicks, have hidden gadgets, and no uniformity like Windows 8 and Ubuntu. I was leaning toward Mint but also considered Free BSD. I couldn't even find reviews for DragonFly. I want a no hassle OS that is as easy and fast to use as the AmigaOS but with more modern software (especially a browser). That's all . Suggestions?

[TeamBlackFox]

Well problem - DragonFly BSD is 64-bit only now. No more i386 releases. So that crosses it off the list.

FreeBSD is great, but its all DIY - you compile the X server from ports or install from pkgng. Once you get it setup, though its very low maintenance. If you're looking for an Amiga style window system I have a few you may like in mind:

AmiWM - Amiga Window Manager Clone: http://xwinman.org/amiwm.php
AfterStep - NeXTSTEP WM Clone: http://xwinman.org/afterstep.php
Enligtenment - The one I use: http://xwinman.org/enlightenment.php

If you're looking for variety of programs you can't beat GNU/Linux, but I don't like it because the entire system feels held together with duct tape. FreeBSD is the choice I'd pick considering your hardware, but you'll have to set everything up in very true UNIX fashion - I like that a lot.

One OS you may enjoy though is OpenSXCE -  a distro of illumos - a Solaris 10 fork:
http://opensxce.org/

Other than that, if I get "Ugly Betty", my Octane, working, then I'd be happy to sell her to you for a decent price - the Octanes are by far the best bang for buck with SGI - IRIX can use Firefox 3.6 currently and while Amigas dominated the low end market, IRIX dominated the high end market.

[matthey]

@TeamBlackFox
Thanks for the suggestions. That gives me some more to investigate. I could upgrade to a cheap duo-core x86_64 box which would pay for itself in electrical costs if I used the system for more than backup purposes but then I would lose my room heater in the winter .

I really like the compactness of the AmigaOS, also, and much of that comes from the 68k. I can debug and disassemble code while being able to read it which is a huge advantage. Optimizing is easier on code that is more human readable also (compilers are not doing their job). No other processor offers these advantages to the extent of the 68k. I am helping with the enhanced 68k Apollo/Phoenix project which is bringing out a new affordable accelerator. I would like to make compiler support easier through 68k ISA enhancements. I am also helping with the vbcc compiler (which also has Amiga roots). Hopefully, we will be able to bring back and enhance the AmigaOS on an enhanced 68k processor and bring more modern software to the Amiga. I actually prefer to stay 32 bit and shrink where most other processors have chosen to go bigger requiring more caches and longer pipelines. I do like a simplified MMU for partial memory protection, cache settings and debugging. Some form of memory expansion and partial program isolation (non-shared parts) may be possible through an extension of the 68k SFC/DFC registers without the overhead of supporting virtual addressing (and without the overhead of 64 bits). Most OS development has been too much of follow the leader and feature focus. Much of the OS innovation (and software innovation) has moved to Android, iOS and BlackBerryOS/QNX (playing catchup but QNX is solid) where some features are no longer expected. Unfortunately, ARM doesn't provide a consistent ISA and the hardware for smart phones and pads can vary considerably.

[TeamBlackFox]

I guess good luck with that. I'm probably NOT going to code for AmigaOS for a few reasons:

No compiler I particularly like for C, and I must use a compiler available for FreeBSD since thats my development system for platforms.
No POSIX compliance
No OpenGL

I'll use it though, definitely as a user. If I were to add the things I want, I'd probably end up with DragonFlyBSD. So instead I stick to FreeBSD and track DFBSD development.

[matthey]

I guess good luck with that. I'm probably NOT going to code for AmigaOS for a few reasons:

No compiler I particularly like for C, and I must use a compiler available for FreeBSD since thats my development system for platforms.

We would like to support vbcc, GNU/GCC and CLANG/LLVM. This requires some support back from the maintainers of the compilers. So far, we have received the most support from vbcc. Vbcc is also the smallest, most modular and has the least number of dependencies. We would like to support them all but we will have to focus efforts also. We will also have to become more popular before we gain more respect. Some are biased against fpga processor support. There has been some effort that has gone into Amiga/AROS or 68k support for all the compilers listed above.

No POSIX compliance

Different levels of POSIX compliance are available through ixemul (GCC targeted), libnix (GCC targeted) and Frank Wille's Posix lib (vbcc targeted). Ixemul basically provides an emulated BSD environment on the Amiga. It makes porting BSD programs very easy but it's not always Amiga friendly and it's big. Libnix and the Posix lib are lighter weight and Amiga friendly but they are not as complete or BSD/Linux compatible. The source code is available so they can be extended as needed.

No OpenGL

The most modern OpenGL/Mesa and 3D support is on AROS but it requires big resources too. It supports hardware rendering through Gallium but there is limited gfx card drivers. There is an old version of Mesa called StormMesa which works well enough with 3D hardware (Warp3D) or software rendering on AmigaOS. Wazp3D can be used for OpenGL/Mesa software rendering in many cases and works on many platforms. UAE can use the underlying OS 3D hardware support. The Amiga needs more work in the area of 3D though.

I'll use it though, definitely as a user. If I were to add the things I want, I'd probably end up with DragonFlyBSD. So instead I stick to FreeBSD and track DFBSD development.

For all the bells and whistles on a workstation/server, DragonFly sounds like it would be super. A modern AmigaOS has possibilities as an efficient personal netbook/notebook/gadget OS. They are very different goals. What we need are options and choices rather than trying to make one OS and processor for everything. All that got us was a security and virus nightmare.

[wawrzon]

The most modern OpenGL/Mesa and 3D support is on AROS but it requires big resources too. It supports hardware rendering through Gallium but there is limited gfx card drivers.

limited? im not sure how limited it is in comparison to other gfx card solutions on amiga like systems, it supports a number of models, namely those that are supported by gallium, and it supports 3d hardware acceleration. sure it might not be optimal, i cant tell, i dont run aros native on x86.

[NovaCoder]

Very nice post, seems to make sense, I think the Amiga OS was ahead of the time in 1987-1990 (maybe even later) and it should be respected for that.  

Of course an OS designed in 85 cannot offer the same features as a modern OS, that would be crazy

It was obviously very advanced when it was first released in 85 but in subsequent years Commodore didn't really do much with it apart from minor enhancements (same as what they did with the hardware!).    Some serious effort was put into 3.9 (maybe 3.5 as well?) but that was still based on the original 85 code base.

Personally I like using Amiga OS 3.9 because of the simplicity and 'retro feel' but it does feel very week in some areas (like file management for example).    With a few minor updates it could be improved to make it even more usable (integrated file management, memory protection etc).

I wouldn't actually want to run AmigaOS 3.x or even 4.x on my PC, maybe I'm in the minority here but I'm quite happy with running Windows on my PC and using my retro hardware for my retro OS fix.   Windows 7 isn't perfect but it's good enough to open my browser and run my Compiler, I don't really need much more from an OS these days.

[matthey]

Of course an OS designed in 85 cannot offer the same features as a modern OS, that would be crazy

It was obviously very advanced when it was first released in 85 but in subsequent years Commodore didn't really do much with it apart from minor enhancements (same as what they did with the hardware!). Some serious effort was put into 3.9 (maybe 3.5 as well?) but that was still based on the original 85 code base.

C= did more to upgrade the AmigaOS than the hardware. AmigaOS 2.x was a major improvement but had it's share of flaws. AmigaOS 3.1 was the best upgrade all the way around. It added a lot of new support as well as key bug fixes. This was the first time that the AmigaOS felt professional. AmigaOS 3.5 and 3.9 were mostly bug fixes and 3rd party add-ons. It was hit and miss as far as improvements. Overall it's good but should have been better.

Personally I like using Amiga OS 3.9 because of the simplicity and 'retro feel' but it does feel very week in some areas (like file management for example).    With a few minor updates it could be improved to make it even more usable (integrated file management, memory protection etc).

Workbench was improved but it could be better. It handles gfx boards and large hard drives better at least. A good file manager is still useful. PeterK's icon.library makes icons fast enough to be tolerable.

I wouldn't actually want to run AmigaOS 3.x or even 4.x on my PC, maybe I'm in the minority here but I'm quite happy with running Windows on my PC and using my retro hardware for my retro OS fix.   Windows 7 isn't perfect but it's good enough to open my browser and run my Compiler, I don't really need much more from an OS these days.

I'm with you. It would seem like heresy. Using UAE on Windows is bad enough .

[Thorham]

I wouldn't actually want to run AmigaOS 3.x or even 4.x on my PC

I  wouldn't run 3.x on my Amiga if I didn't have to. 68020+ can do SO much better than AOS.

[KimmoK]

Amiga OS 3.9 already runs everywhere. So?

It only is limited to past and requires JIT for productive use.

(running AOS3.9 on emulator on x86 is cost efficient way to use 68k Amiga on high res display, on laptop and with some 1Ghz 68060, other than that, not interested. + MorphOS on PowerBook already covers my mobile 68k/AmigaOS needs. (and  I also have AOS4 system))

[agami]

This is an extremely valuable question and not enough people ponder it.
The Amiga OS outside of the Amiga Hardware still has a lot of unique things for which it would be cool to have it be able to run on more modern hardware.

History
I have since before the Amiga and up until now used and continue to use many different computing platforms and operating systems, both professionally and out of my own curiosity.
In the late '90s just as Gateway purchased the rights to the Amiga I wrote them an extensive analysis on why they should give up on the hardware game and focus only on the OS. I was shot down by Gateway/Amiga Inc. as they believed that the OS had now future outside the custom hardware. I bet Carl Sassenrath would also disagree with them.

The Little OS that Could
Without a doubt the Amiga OS, true to its goal, is the most user centric and the most user friendly OS ever developed.
This has nothing to do with where the widgets are placed in relation to the window, the small footprint, or the inclusion of pre-emptive multitasking. It is about the paradigms.
The OS paradigms that are at the core of the Amiga OS are in certain ways more relevant today than they were in the late '80s and early '90s. And that makes it doubly tragic that all of the post OS3.x forks are missing this important point.

At the end of the day there aren't many unique OS paradigms in existence today. There is the Unix paradigm, the Mac OS (pre-OS X) and Windows paradigm, there is an assortment of experimental OS paradigms that will most likely never get their "moment in the Sun", and then there is the Amiga OS paradigm.
And yes, there are some superficial similarities between all of these which are there for a bunch of different reasons and have a lot to do with convention.

A really good OS strikes a balance between the user and the developer. As an OS is primarily an environment in which applications are run, it would behove itself to make it easy for application developers to create applications for it. Make it easy for developers to write the kinds of software users would like to use.

Windows is terrible at this. Mac OS X is slightly better only because of the energy that Apple puts in to establish the benchmark and by controlling the tools. Don't even get me started on Unix and Linux.
Amiga OS is the only OS I have had the pleasure of knowing that did this well, without the overbearing control, and in the late '80s and early '90s for crying out loud.
And yes, features such as memory protection, and multiuser environments and better ACLs could be implemented if there was funding.
Amiga OS could be like many of the other OSs, but other OSs have tried to be more Amiga like and failed. There is something in that, do you think?

[OlafS3]

while it is not a priority this side is being optimized and bugfixed as well.

Nice to hear :-)

perhaps you could do a step-by-step manual how to install Aros 68k (nightly) on real hardware. I was often asked but I cannot help there.

[OlafS3]

I'll give you that compactness can be nice.

AmigaOS is my favourite graphical OS. If it had a more UNIX-like commandline and had some of the features I mentioned it missing it'd be pretty much the perfect OS.

Since it won't likely get those for a long time, my bet is thrown in the hat with DragonFly BSD. What Matt Dillon promises is what people want of AmigaOS with the UNIX philosophy behind it. ( Once Wayland becomes BSD friendly we'll be able to ditch horrible X11 )

In my distribution there are many commands that are identical to unix world. Amiga 68k developers were heavy inspired by the unix world.

[wawrzon]

Nice to hear :-)

perhaps you could do a step-by-step manual how to install Aros 68k (nightly) on real hardware. I was often asked but I cannot help there.

there is one instruction for instance here, if you scroll down a little:
http://en.wikibooks.org/wiki/Aros/Platforms/68k_support
must have been literally copied from my post, since it ends with "thats all folks" (without apostroph).

now it might not be exactly up to date as things change all the time, for instance you need additionally edit the startup sequence now, but roughtly it still stands i guess.

[vxm]

C= did more to upgrade the AmigaOS than the hardware. AmigaOS 2.x was a major improvement but had it's share of flaws. AmigaOS 3.1 was the best upgrade all the way around. It added a lot of new support as well as key bug fixes. This was the first time that the AmigaOS felt professional.

Yes, the evolution of the AmigaOS under the auspices of Commodore was notable.
Maintain a certain level of backward compatibility both hardware and software
while offering innovations (datatypes, rtg, tags, etc) was a real challenge.

[TeamBlackFox]

Seems like Agami has some of the right ideas but I want to refute one in particular:

> Amiga OS is the only OS I have had the pleasure of knowing that did this well, without the overbearing control

I'll have to disagree here. AmigaOS is HEAVILY biased towards the user. Consider that UNIX was developed in the 1970s, and with it and Berkeley UNIX you get the following things Amiga still lacks:

Memory Protection
Privilege separation
Multi-user support

Plus there is a balance between the user and the developer in UNIX. Just most people are too dense to see it. Let me explain, but first:

> Don't even get me started on Unix and Linux.

Please don't make the mistake of blobbing these together. GNU/Linux is horribly biased to developers, and it promotes bad, bloated, lazy code.

Modern UNIX descendants like System V and BSD are primarily hindered by the horrible display server that is X11. Each variant of BSD and System V variant addressed this differently:

Sun developed NeWS, a Display Postscript variant, for SunOS, their Berkeley UNIX derived OS ( Berkeley UNIX refers to historical versions of BSD here ) but it failed horribly in the market due to X11 being very entrenched

NeXT Computers developed their variant of Display Postscript. Even though NeXTSTEP is not a true Berkeley UNIX derivative ( Based off Mach, not UNIX ) it does maintain UNIX compatibility, and their Display Postscript server technology was passed into OS X as Quartz.

SGI developed XSGI, their variant of X11 which addressed its flaws by optimising it for SGI hardware and stripping out what wasn't needed.

The others adopted X11 and dealt with the shortcomings. AMIX being a System V derivative was among these, notably.

However with Wayland under development we should see all the inherent flaws of X11 be corrected. Wayland is a proper display protocol which doesn't treat all hardware like a big dumb framebuffer ( What X11 does without the hacks like DRI and such that people have been working on )

The reason I say UNIX proper can balance user and developer focus is simple:

Its well known that UNIX itself is one of the most developer friendly OSes of all time.

Once the shortcomings of X11 are gone we are left with only one major issue - a lack of a standardised toolkit. That can be addressed down the line, for now ditching X11 is by far the most imperative issue, its almost 20 years late after all.

The biggest issue in my opinion today is that most consumers are morons and are afraid of working in the console. Thats why I point newbies to UNIX to FISH, the Friendly Interactive Shell. Useless for scripting, but really assists new users by being actually helpful and interactive rather than biased towards developers. You throw together Wayland X Enlightenment X FISH and most users after the initial learning curve won't have any issues.

Enlightenment is my choice of GUI due to its minimalism, yet simplicity of use while being eye appealing and not a resource hog.

I am far from a critic of Amiga, I'm an advocate actually, but I think its best chance of not fading into obscurity relies on the promotion of DragonFlyBSD.

I rest my case in the matter at this point. Take it however you will.