@matthey

Quote:

The AmigaOS uses C and there isn't much change after C99 and C11.

Yes, C17 was a bugfix release -- no new features. It's nice that they did this. I recently moved from C11 to C17, and will probably stay with it for a long time.

_________________
"Unix is supposed to fix that." -- Jay Miner

@agami

Quote:

I remember reading similar things back in the day, but when I corresponded with Ben several years ago, he was adamant that there is no legal/contractual prohibition of porting AmigaOS to x86.

That's interesting. It's been years since and any original contracts would be made with Amiga Inc. Now more complicated as they dissolved and Cloanto becoming involved.

I find it hard to find now, and I've located it in the past, but I think this is it and quick make it a sticky post now!

http://amigaworld.net/modules/newbb/viewtopic.php?topic_id=18848&forum=2&start=40&viewmode=flat&order=0#289728

Quote:

What I think he omitted to say, and in truth it wasn't germane, is that a port of AmigaOS to x86 would contractually not be able to be branded as AmigaOS 4 or 5, or any number in the series. Potentially not even AmigaOS.

That doesn't sound very useful. There is less in common with an Amiga and an AmigaOne compared to AmigaOS and OS4. I think for for an AmigaOS 5 it would make sense, but that was going to be based on AmigaDE and those AmgaObject ideas. Which didn't eventuate but an AmigaOS 6 is like x86 so could be a link there.

Quote:

Furthermore, Ben said that they had at one point looked into an x86 port, and it was just too large of an undertaking. Which makes sense when you sneak a peek behind the curtain and see just how small the Hyperion operation is.

Yeah, two programmers. Correction, no programmers, and contracted to two programmers. It ruined their life, apparently.

Logically, before OS4 could be ported to x86, it would need to be ported to little endian first. That's the first step I see. Of course, at that point you have backwards compatibility issues. Do you drop it or sandbox it? It would be inconceivable to drop support. So 68K must be supported. Though AROS has struggled for transparent 68K support, while Amithlon supported a native x86 binary making calls into a 68K OS, it surely must be possible to support running 68k apps (and now ppc apps) transparently. Likely a legacy emulation layer would be needed that hosts a legacy Exec and libraries that cleanly translates to native OS calls. Similar to how 68K code can jump into PPC native.

In the real early days of the OS4 port, I recall a status update, of an OS4 mini shell. I'm vague on details but it must have been when they were detaching from Amiga hardware and running it solo on the AmigaOne. I'm sure I also read that they had compiled base code for x86 and had the OS4 mini shell running on x86 as well but I don't know if any evidence is left online.

@Karlos

Yes I wrote The Maestrix. I even managed to email Richard Korber and got permission to use a derivative of his work. Yeah sorry about the bug. It bugged me too lol! It turned out I had missed a device reply as they were tied to one signal so if two came back at once it missed it so it caused the desynchronisation to occur. Funny as two data messages could come in at once and be processed. I had been through the code since and noted problem areas I set about fixing years ago but you can thank Helge for encouraging me to fix it. And I finally had a bug free release early last year! I also updated stuff and shared a guide with what MED mix buffer means. Then, late last year, managed to port it to PPC with my first ever MorphOS release. Again, Helge encouraged me to do it, though some would say badgered. But it was my plan to make a native version along.

Funny thing is, when I first wrote it, I didn't expect it to last so long or be relevant to OS4 still in this day. It was just meant to be a temporary project as at the time OctaMED was going to be ported to OS4. Well, he we are still waiting, and so The Maestrix still remains to be relevant. Plus, I found what froze MED on MOS, so wrote a patch called OctaMEDIC. Oh no, another theme...

Quote:

Looking at this include section and remembering how much of the code is lost forever gave me a sad:

That's a shame. Also see a 2006. Must have been a good year for Amiga development. Can see the effort put into that. What happened to the files?

However, 0xABADCAFE. Haha, funny.

@Hypex

It's cool that you've made OctaMED more accessible to everyone with those tools. Outside of an IDE or shell, I've almost certainly spent more time in OctaMED than in any other Amiga application. Whenever I do manage to find time to spin up AmigaOS, it's usually the very first place I go after it boots. I used to be a pretty prolific creator of unfinished tunes, lol.

Quote:

That's a shame. Also see a 2006. Must have been a good year for Amiga development. Can see the effort put into that. What happened to the files? However, 0xABADCAFE. Haha, funny

That's a long story. The GitHub code is a surviving fragment of the second iteration of a project I'd created previously. The first iteration only ran on Amiga OS 3 (building in storm C) and Win32 (watcom). It was a framework that allowed me to quickly create applications that had graphics, audio etc. It was very much a minimally C++ version then, with no support for threads, didn't use exceptions, etc.

The second version was more ambitious. Using lessons learned I manages to get working builds for pretty much all the targets in that list. All at varying stages of completion. I was a bit lax with backups but set up a local SVN repo on my indestructible Linux box. Turns out it wasn't as indestructible as I thought and suffered a total disk failure. I still have the disk and should really get it recovered but it's not a justifiable expense at the moment.

This second version was an effort to maintain, spanning no fewer than 3 different GCC versions from 2.95.3 to 5. It is full of odd macros to deal with this. For example, by putting an exception throw into a C function with C linkage, 2.95.3 produced code about 35% smaller with no apparent loss of functionality. Similarly, the right declaration at the right time produced much less bloat where RTTI was used (which was used to report exception names where caught).

This was the version that ran aground on OS4 when the thread library didn't handle exceptions correctly, despite working in 3.x and Linux variants. The MorphOS and AROS targets never got far enough to test with that.

It was a fun project for me because I love reinventing wheels. I'm that guy that plays a game and is more interested in how it works than the game itself. There were also some cool, evil experimental hacks. For example, you could enable low level exceptions in 3.x and get a zero overhead (when it doesn't throw) integer divide by zero exception. This worked by using an asm handler that changed the return address from the hardware exception to a function that then threw the zero divide. The intercept was here https://github.com/0xABADCAFE/ancient-amigaos-exng2/blob/69c07dc1fcc215c81c2bf5ea7e25a83ca6a1d9e4/libsource/platforms/amigaos3_68k/systemlib/thread_native_asm.asm#L36

A working proof of concept (the binary is in the same location) is here
https://github.com/0xABADCAFE/ancient-amigaos-exng2/blob/69c07dc1fcc215c81c2bf5ea7e25a83ca6a1d9e4/projects/test/exceptions/throw.cpp#L20

Looks like my const correctness wasn't as keen then lol

Yep, those were days of constant exploration and invention for the fun of it... Happy times.

Last edited by Karlos on 05-May-2022 at 10:13 AM.

_________________
Doing stupid things for fun...

What gives us ppc?
Amiga ppc is Amiga that Commodore made if not bankrupt.
Something native, binary and source compatible with old 68k software.
Something as fast and as comfortable as at least cheap pc from win95 era.
Something as fast in 2D and 3D as at least playstation 1.
Something that gvb and natami/apollo/vampire team promise but never deliver.
Something that 68k never was.
060 still has not fast 2D ad 3D.
Vampire still has not 3D and still has not compatible fpu and mmu.
Only Amiga ppc has that all.
Amiga ppc is something like wonderful super classic.

@ppcamiga1

It's impressive just how much incorrect nonsense can be packed into so few lines.

Quote:

What gives us ppc?

Phase5 Digital Products.

Quote:

Amiga ppc is Amiga that Commodore made if not bankrupt.

No it isn't. That would have been the Hombre chipset, which was designed around Hewlett Packard's PA-RISC. It had no relation to Power PC. It also had no relation to OCS, ECS, AGA or AAA. It was a completely new design.

Quote:

Something native, binary and source compatible with old 68k software.

No. There is no binary compatibility between 68K and PowerPC, or the PA-RISC of the Hombre. Both depend on emulating 680x0 for any degree of backwards software compatibility.

Quote:

Something as fast and as comfortable as at least cheap pc from win95 era.

You are of course joking. For the cost of a CyberstormPPC you could have bought a complete Win95 era PC.

Quote:

Something as fast in 2D and 3D as at least playstation 1.

That's not an especially high barrier. A 68060 with Permedia2 can push more triangles and at a higher resolution and colour depth.

Quote:

Something that gvb and natami/apollo/vampire team promise but never deliver.

They have never promised a PowerPC solution, so it's only natural they'd not deliver one.

Quote:

Something that 68k never was.

Well that's true. They are different processor lines.

Quote:

060 still has not fast 2D ad 3D.

For what? Context is important. 68060, especially with a decent RTG card is more than fast enough for the sort of 2D and 3D that were relevant at the time the 68060 was released.

Quote:

Vampire still has not 3D and still has not compatible fpu and mmu.

Not much 68K amiga software especially depends on MMU. The FPU emulation is or was imprecise for productivity software but it's an FPGA solution so that's fixable with an update.

Quote:

Only Amiga ppc has that all.

Not even close. I had to use UAE to run legacy chipset banging software on my A1. That's no different to running UAE on a PC. Actually, no wait, it is different. The PC build was faster and had more options and was more compatible.

Quote:

Amiga ppc is something like wonderful super classic.

No it isn't. You want a "super classic", just run UAE on a current generation PC. Amiga "PPC" is the just about the least cost effective way of having a "super classic" that I can think of.

For system friendly Amiga applications, PPC is a decent enough experience but the simple, unavoidable fact of the matter is that the hardware line has ran it's course. The PowerPC is less relevant to the desktop computing segment now than 68K was when PowerPC was introduced. It doesn't matter at all what you think of AMD/Intel, they took over the entire segment. It's only recently that they face any serious competition from ARM / Apple, but since the latter will only see the light of day in apple products, I don't think they're cacking their pant over it yet.

_________________
Doing stupid things for fun...

@Karlos

Quote:

It's impressive just how much incorrect nonsense can be packed into so few lines.

It's his super-power.

_________________
All the way, with 68k