What's the deal with NG Amiga OS's?

[MartinW]

OK, right up front, for the love of god, please don't let this degrade into a thread about technicalities over hardware. Even though I used to love fiddling with hardware, have tried to get into FPGA's and have built my own GBA, at the moment the hardware arguments over technicalities and instruction sets etc. bore me to tears (apologies to those that love that stuff - I still read and for the most part understand it, I just can't feel it right now).

Over the last few weeks I've played a bunch with Aros and Morphos. I've built Aros from scratch and hope to take it further and I've tried the latest 3.10 version of Morphos and was really rather impressed*.

Two things puzzle me however.

1. I way be wrong, but I believe that both Aros and Morphos claim to be compatible at API level with AOS 3.1 - that given, how come we can't just compile stuff from one to the other? Or maybe we can? Maybe with some macros to convert API function names?
2. What is it with these OS's that make them 'Next Gen' Amiga OS-Like. I understand the 'feel' of the OS, but can I run my 68k software somehow other than just emulation? If I have to resort to emulation then I have a perfectly capable Skylake Windows system to do that. Not to mention a MacBook Pro that cost me £2.5k that can run OS3.9 and OS4FE under emulation like a champ.

I'm kind of struggling on what makes these Next Gen and / or the desired way forward??

Beyond the "I quite like this" I'm a bit stuck what I should be getting so excited about once I've got each up and running and skinned / themed them to my liking?

But I am in limbo a bit at the moment and sort of looking for a new hobby. Like I say, I'm kind of tired of the hardware side. I have a tonne of stuff here. Various computers and a LOT of arcade hardware, not to mention that my eyesight isn't what it was and I have to resort to huge magnifying glasses in order to solder. I very much feel the urge to get into a hobby that get's creative with what's around me rather than amassing more. But I don't have a specific software project in mind. My job kind of sucks the life out of me with having bright ideas on that front.

*Impressed I may be, but I'm STILL reluctant to shell out the asking price for the license when it is locked to hardware that is known to be a bit rubbish (ATI Radeon with 32MB of video RAM), especially now it is no secret that the Morphos team are moving towards targeting x86 at some point and presumably then my money would be spent on already obsolete hardware.

[swinkamor12]

Originally Posted by MartinW

I way be wrong, but I believe that both Aros and Morphos claim to be compatible at API level with AOS 3.1 - that given, how come we can't just compile stuff from one to the other? Or maybe we can? Maybe with some macros to convert API function names?

AROS is not compatible at API level with AOS 3.1- funtions form os exist but they are not tested and do ditterent things than original AOS 3.1.
Amiga NG is compatible, functions form os exists, they are tested and do what they are do on 3.x;
Plus You can use 68k library in ppc code.

Originally Posted by MartinW

1. What is it with these OS's that make them 'Next Gen' Amiga OS-Like. I understand the 'feel' of the OS, but can I run my 68k software somehow other than just emulation? If I have to resort to emulation then I have a perfectly capable Skylake Windows system to do that. Not to mention a MacBook Pro that cost me £2.5k that can run OS3.9 and OS4FE under emulation like a champ.

On pc emulated is whole system.
On ppc only program code is emulated, os is not emulated and system calls run full speed on ppc.
On mac mini 68k software run faster than on my i7.
If someone have old pc, old mac mini will be cheaper than upgrade to for ex i7 to and give better results.

[ajk]

-- but can I run my 68k software somehow other than just emulation?

What other option could there be than emulation, if the CPU architecture is different? The software can of course be recompiled for the new system (ideally without any changes, or only minor ones) if the sources are available.

The 68k emulation (at least in the case of OS4 and MOS) is built into the system, so you can simply run a 68k binary as you would a native one. The process is quite transparent.

You only need to run an actual emulator program (UAE) if you also need to access the chipset (again something that is missing from the new systems on a hardware level, so it has to be emulated). This would be most games, as most of them don't use the system libraries for accessing the hardware (for obvious performance reasons).

[Minuous]

It seems anything that runs on something other than 680x0 is referred to as "NG".

However it's rather meaningless because AROS, for example, is supposed to be one of these "NG" operating systems, yet nearly two decades later they haven't even implemented OS3.9 levels of functionality, let alone OS4.0/4.1.

It would be like if I wrote a clone of Windows 95 and called it the "Next Generation Windows OS".

MorphOS is not much better in this regard either.

[Daedalus]

Originally Posted by MartinW

[*]I way be wrong, but I believe that both Aros and Morphos claim to be compatible at API level with AOS 3.1 - that given, how come we can't just compile stuff from one to the other? Or maybe we can? Maybe with some macros to convert API function names?

They (and OS4) have 3.1-compatible APIs, and that means that most 3.1-targeted software can be recompiled for them with little or no changes to the source. The problem comes from the additions and updates to the API that make it a superset of the 3.1 API. These new calls, new interfaces and new libraries are all unique to each "NG" OS, which means that recompiling from one to another isn't as easy. If a coder simply sticks to the 3.1 API however, that should be recompilable across all OS variants.

[*]What is it with these OS's that make them 'Next Gen' Amiga OS-Like. I understand the 'feel' of the OS, but can I run my 68k software somehow other than just emulation? If I have to resort to emulation then I have a perfectly capable Skylake Windows system to do that. Not to mention a MacBook Pro that cost me £2.5k that can run OS3.9 and OS4FE under emulation like a champ.

Part of what makes them "Next Gen" is that they're simply a new iteration of the OS. Back in the day, Workbench 2.0 could have been considered "Next-Gen" in relation to 1.3, the A1200 could have been considered "Next-Gen" relative to the 600, and so on. Next-Gen is just a fashionable phrase in more recent times to describe the next version of something. As for the NG OSes, like I said above, all have added lots of new features over and above the 3.1 function set. Each OS has a different set of new features as they all have slightly different paths that they follow, but all three have things like support for "modern" hardware, native 24-bit colour, USB support and so on. Then you have things like expanded APIs in almost every area of the OS, adding new calls to help the coder, handle new hardware, or just replace tired, old techniques.

When it comes to emulation, MorphOS and OS4 have an inline 68K emulator that works very well to run system-friendly 68K software while using the NG API. This is much faster than emulating the entire machine, and means much better integration - 68K programs don't look or feel any different to PPC programs, and aren't isolated from the system, which allows things like seamless ARexx usage and drag'n'drop between 68K and PPC applications. Think Rosetta on OSX Leopard. It's a very, very different experience to running a virtual machine.

I'm kind of struggling on what makes these Next Gen and / or the desired way forward??

Each OS represents one possible way forward for the Amiga OS. For a lot of people, they don't really want the OS to be brought forward, or don't care enough about it either way as it's not their main concern, and that's fine too. But some people like having a glimpse of what the future of the OS might have been, had Commodore's story been a little different.

Beyond the "I quite like this" I'm a bit stuck what I should be getting so excited about once I've got each up and running and skinned / themed them to my liking?

If using system-friendly software isn't a big thing for you then it's possible that there isn't much beyond playing about with themes to entertain you. But for me personally, using an NG OS feels like the next evolution from what was my main computer in Uni around the turn of the century: a massively expanded A1200 with an 060, graphics card and OS 3.9. For example, for me coding Amiga software, anti-aliased font rendering makes it a much more comfortable experience for me, even if I'm coding 68K stuff. I then cringe a little whenever I move back to 3.9 for testing or whatever reason. And there are plenty of other minor details that I miss when moving back that, each taken on their own doesn't seem to be a big deal, but when taken together all add up to a much upgraded experience.

But I am in limbo a bit at the moment and sort of looking for a new hobby. Like I say, I'm kind of tired of the hardware side. I have a tonne of stuff here. Various computers and a LOT of arcade hardware, not to mention that my eyesight isn't what it was and I have to resort to huge magnifying glasses in order to solder. I very much feel the urge to get into a hobby that get's creative with what's around me rather than amassing more. But I don't have a specific software project in mind. My job kind of sucks the life out of me with having bright ideas on that front.

The whole NG thing is a move *away* from custom hardware towards more off-the-shelf gear, so there's less soldering, less dependence on custom chips, and so on. Yes, there are custom motherboards in some cases, but they still take standard format PCI and PCIe cards. NG systems are all about the software end of things.

Perhaps you could help out with the AROS project, or porting open-source frameworks and applications to any of the NG systems.

*Impressed I may be, but I'm STILL reluctant to shell out the asking price for the license when it is locked to hardware that is known to be a bit rubbish (ATI Radeon with 32MB of video RAM), especially now it is no secret that the Morphos team are moving towards targeting x86 at some point and presumably then my money would be spent on already obsolete hardware.

Why is it only then becoming obsolete? Surely it was obsolete before you even started? after all, it's over a decade since Apple dropped PPC. If it works, it will continue to work equally as well if/when an x86 version is released. And that will be many years away at least. The move to x86 has the potential to drop all compatibility with 68K stuff at an API level, keeping 68K software sandboxed in virtual environments for compatibility. That being the case, I can still see demand for PPC hardware, so while you might not want that old Mac that has suddenly become obsolete, others might, and you should be able to sell it (with the licence) and maybe cover the cost of an x86 licence that way.

Anyway, the bottom line is that you should choose a hobby that makes you happy. If you genuinely don't see any appeal to the whole NG area, even after spending some time with them, then that's fine. Stick with classic, which is still well catered for in both hardware and software terms.

[AMIGASYSTEM]

Originally Posted by Minuous

It seems anything that runs on something other than 680x0 is referred to as "NG".

However it's rather meaningless because AROS, for example, is supposed to be one of these "NG" operating systems, yet nearly two decades later they haven't even implemented OS3.9 levels of functionality, let alone OS4.0/4.1.

Some parts of 3.5 and 3.9 are already implemented, remember that AROS is an Open Source system and can not use commercial material.

http://aros.sourceforge.net/document...rs/roadmap.php

[MartinW]

Originally Posted by swinkamor12

On mac mini 68k software run faster than on my i7.

OK, that's interesting indeed. So I think it may be time to register MorphOS (if that's what you're referring to?) to get rid of the time limit and dig a bit deeper into this then.

I have a great 68060/100Mhz system with PII++ card but the monitor and screens are always still a pain. The Mac Mini goes so much easier with a modern LCD.

@Daedalus - a great response with a lot of detail, thanks. I'm going to re-read it in a bit but I think I mostly get the gist of it. In particular I think I now understand what the different NG's do and don't do in regard to the APIs and if I understood then if I took something (or created something) that was 100% compatible with the OS API calls from (say) 3.1 then it should either just build or not require too much in the way of changing to get to build on any of the NG's?

Just to go back to the Mac mini and MorphOS again, I think my biggest issue is that I have an A1200 and a GBA if I want to purely play games so while the Mini is a great machine I have a tendency to steer towards things that the other two machines can't do and then I start to hit the limits of that 32MB of video ram. MorphOS feels to me like it could do with some better hardware than the Mini but then we hit the expensive hardware problem again.

I'm very likely to jump on an A1222 assuming they see the light of day. I do like OS4.1 (and did buy a copy), I just don't like running it emulated.

[Gorf]

I think one problem is the term "Next/New Generation".
All three are either vintage and/or outdated themselves from a todays perspective or never reached a "production level" status, like AROS.

Trying to stay more or less compatible to the original API and trying to offer a experience that is very close to that of the classic Amiga is both: goal and burden at the same time.

that's why the other thread (the boring API, ISA and so one discussion) tried to rise the question what a modern system might look like and if that would be still Amigaoid.
Probably not, as everybody has his/her very own definition what makes an Amiga and what features are essential and what not....

From the mentioned trio AOS4, MOS and AROS, i think the last one has the biggest potential, as it can run native (and hosted) on modern hardware and different architectures ...
Sadly it is missing developers and general interest and the overall experience can be quite frustrating - for users and developers alike

A couple of years ago there was a big AROS bug-fixing and code-cleanup effort, that improved usability greatly and was a success ... but sadly not a lasting thing.
I hope this can be repeated for ABIv1 in not so far future.

[MartinW]

I've not done much with any of these yet as you've seen, but something from my experience with MOS this weekend that would benefit Aros greatly is the development setup. They have a nice IDE now called Flow Studio and I can't stress enough how good it felt to just install the latest SDK, bring up that IDE start a new project and 2 minutes later there is a fully working Hello World complete with Makefile and debugger support. I didn't do anything much apart from a few mouse clicks and type the lines of code.

That has to be a win surely?

[Gorf]

At least definitive guide how to setup a working development environment for AROS would be great. Along with overall better documentation and tutorials.

And than there is the whole ABIv0 vs. ABIv1 disaster. Many things are fixed/improved in ABIv1 but back porting has a very low priority ... most devs are on v1 anyways ... so v0 (were most or all distros are) is somewhat left out in the rain.

So just go on AROS-people: make the jump! set v0 as "deprecated" and move everything to v1 - experimental stuff can go in a new v2.

But there seems to be nobody who could even make such a decision ....

[jPV]

Originally Posted by MartinW

Just to go back to the Mac mini and MorphOS again, I think my biggest issue is that I have an A1200 and a GBA if I want to purely play games so while the Mini is a great machine I have a tendency to steer towards things that the other two machines can't do and then I start to hit the limits of that 32MB of video ram. MorphOS feels to me like it could do with some better hardware than the Mini but then we hit the expensive hardware problem again.

I wouldn't say you hit the expensive hw problem with MorphOS that easily Practically any of the supported Macs can be found for under 100e nowadays, even powerful G5 machines. There are many cheap options for laptops, big towers, etc which do have more graphics memory, if that's your issue. Or if you don't like laptops or big work-horse machines, hunt a 1.5GHz Mac mini (the silent upgrade model) and that has 64MB of vmem. It's a nice MorphOS machine which doesn't take too much space

[MartinW]

Thats a fair point. And I have to say, another reason for not registering this G4 mini yet. I'm tempted to go off and see if I can pick up a G5 tower. There's a company locally doing a refurbished one with a warranty but it's bit overpriced (not much, but it's certainly nearly double the 100€ mark). I could always end up modding the case and using it for my Hackintosh

I have to say, I'm more and more impressed by MOS now. I've just brought it downstairs and connected it to my main monitor which is a 2560x1080 ultra-wide. So I have MOS running 'picture-by-picture' next to my main laptop and using synergy as the mouse / keyboard client so only one keyboard / mouse. That's pretty slick I have to say.

[trixster]

You can pick up a G5 2.7ghz monster for about £70 if you search around. That's what I did. It's a cheap way to experiment with MOS.

[MartinW]

Are the G5 towers very noisy? Looks like there's a lot of fans in there!

[trixster]

Not especially. It's quieter when running OSX as the fan throttling works correctly. Under MOS it's a little louder as I don't think the OS quite knows what to do with the fans with only one CPU working. It's not loud, but it's not anywhere close to silent.

[MartinW]

OK, that makes sense. I thought the G4 mini was quite quiet until I shut it down :P

[jPV]

Originally Posted by MartinW

I'm tempted to go off and see if I can pick up a G5 tower. There's a company locally doing a refurbished one with a warranty but it's bit overpriced (not much, but it's certainly nearly double the 100€ mark).

I personally would recommend to pay some extra if you know that machine will be in good condition, because we know that it can be in good use for years in our hobby, it's not going to get obsoleted any time soon. Even if you pay double, it'll be really cheap in Amiga terms

I have to say, I'm more and more impressed by MOS now. I've just brought it downstairs and connected it to my main monitor which is a 2560x1080 ultra-wide. So I have MOS running 'picture-by-picture' next to my main laptop and using synergy as the mouse / keyboard client so only one keyboard / mouse. That's pretty slick I have to say.

Great to hear this kind of comments for a change, and I fully agree We "NG" users, and our "NG" operating systems, usually just get bashed in this particular forum There are so many great things on "NG" systems if people would look things from a different angle sometime and actually check what we have now.

[MartinW]

Originally Posted by jPV

I personally would recommend to pay some extra if you know that machine will be in good condition, because we know that it can be in good use for years in our hobby, it's not going to get obsoleted any time soon. Even if you pay double, it'll be really cheap in Amiga terms

OK, heard back from the seller and that's got an NV5200 card in it. Shame because by the time I add the price of a reasonable Radeon and the price of the OS, sure it's still cheap compared to X5000 etc., but I do feel it's getting a bit high. You never know, I might change my mind.

For the moment I think I'll stick with the Mini and just keep my eye out. Key is to get into doing something with it first I think.

Originally Posted by jPV

Great to hear this kind of comments for a change, and I fully agree We "NG" users, and our "NG" operating systems, usually just get bashed in this particular forum There are so many great things on "NG" systems if people would look things from a different angle sometime and actually check what we have now.

Well, credit where it's due. I first tried MOS quite some years ago although I don't remember what version it was then and the experience was less than stellar. Back then wireless didn't work on the mini, and if I recall the experience generally wasn't too clever.

I tried 3.9 last year and having working wireless was a major improvement (I may have tried a version in between too?). This time I themed it to make it more Amiga-like (don't shoot me!) but ultimately didn't do anything with it.

3.10 seems to be a lot more polished. I don't think I've had any problems with it at all and like I say, Flow Studio is very nice icing on the cake so this time I'm going to try and stick with it.

[MartinW]

Originally Posted by Gorf

At least definitive guide how to setup a working development environment for AROS would be great. Along with overall better documentation and tutorials.

Couldn't agree more with this. I spent HOURS last night trying to get a version that compiles so I can get the cross-compilers. I tried from the git mirror and 'out of the box' I couldn't work out how to compile contrib. After it dawned on me that it was because it was outside of the main build tree I copied the sources to a temporary place first in the right order and then I was at least able to build.

But it would fail. It looks like it was down to the combination of binutils and GCC version. Ultimately 'system' build with gcc 6.3.0 and the default binutils. It wasn't until after I thought to look in the official build logs to see what their configure setup was. For some reason if I try to compile with the binutils they are using, it fails to build. I guess there's something wrong on my end. And the default GCC (4.x) is fundamentally broken with the version of texinfo that installs on Ubuntu.

Then there's contrib. Damned if I can get that to build. It fails in one of the supporting port packages. Both on i386 and x64. Don't know why yet because it doesn't on the build server.

Would be so much easier if someone could just supply pre-built cross compilers.

The good news is that what did build of contrib works better than last time I tried it. No crashes this time except in the 'Galaxy' demo.

[Korodny]

Originally Posted by MartinW

I'm kind of struggling on what makes these Next Gen and / or the desired way forward??

These projects were started about two decades ago. Back then, there was an actual community of Amiga desktop users who used their machines they way others used their PC: read mail and usenet, browse some web sites, write a document, listen to some music... I still remember the joy many of us felt when Amithlon got released - finally, you had some decent processing power available...

That's the people OS4, MorphOS and AROS were targetting. But these projects came too late, delivered too little or got stuck in bullshit legal battles - the target audience simply vanished in the meantime and the niche one of these system could have maybe occupied is gone.

Beyond the "I quite like this" I'm a bit stuck what I should be getting so excited about once I've got each up and running and skinned / themed them to my liking?

That's all there is to it, these days - plus the joy of owning a exotic piece of hardware/software.

OS4 is dead, AROS was never really alive and MorphOS is making some impressive progress (there are lots of hints they're working on a x86/x64 port), but software development for it died down years ago as well.