Is anyone actually testing ABIv1?

[aGGreSSor]

Hi. I tried building a Ubuntu-based virtual machine which easily allow to compile abiv1-64 executables from Linux. Look for it in Icaros Desktop's website.

Yes, I know. Thank you, but this is not a toolchain and does not integrate into my existing development environment, so I put off testing this method for now. 

To be honest, though, the high curiosity about Icaros 64 was followed by the total lack of interest by 3rd party coders, who, with a couple of exceptions, completely avoided compiling things for 64bit architecture.

I'm sure the reasons are the same as mine.

Moreover, most of the programs you see on the Archives, although compiled for x64-aros, are NOT compatible anymore and nobody cares about this, neither aros developers, nor applications ones.

It's much worse. Most of these programs are'nt compiled at all for x64-aros.   For example. Yes, I understand why it happened, but use "file" in Linux..

So I wonder what the hell we're doing here. I don't like spending my time, money and efforts on something nobody cares about. So I turned back to my old cheer ABIv0 32-bit icaros and put the 64 bit version in the fridge. Better times will come, to bring it back to attention. At least I still have some real USER that enjoys what I do.

This is expected and logical

You can't attract developers if you continue repeating they will need to recompile their software over and over again.
You can't even attract the one-time casual dev, if AROS 64's gcc doesnt even work
and you definitely can't attract USERS, if you don't bring APPLICATIONS.

Unfortunately, this is generally unrealistic, a certain number of man-hours were spent on 1412 packages in the archive (this is very little, but all that we could). Even if you consider that not all of this is an applications, a lot still needs attention. Open software culture is not typical on Amiga and this has only begun to change in recent years. Therefore, for most applications there is no source code, including those that are in the archive. Proprietary applications for m68k-Amiga written in assembler for AROS are generally lost. You can recompile two dozen applications that will not be enough for work. Before that you need to recompile the libraries, for example, SDL, SSL, etc. Here there are no people versed in cryptography, I myself passed only next to this topic. Therefore, we will forget about cryptography in AROS applications for now. Along with this and about applications that require secure network connections.

All have a SDL libraries of varying degrees of curvature (so no one uploads their versions). A somehow working ncurses relies on SDL, for example. It takes a lot of patches for this to work in an Amiga environment, which is low-level work. You can't just take and compile. Although as a first approximation with POSIX everything is good, in practice, elementary things are missing in ABIv0 (!). For example, it's good if you know about this repository. Although this solution to only a few problems is incompatible with future AROS. We are also limited by memory from above, do not forget about it. And after that we say that we need to recompile everything, but the stable version of ABIv1 is being postponed for now.

[paolone]

Unfortunately, this is generally unrealistic, a certain number of man-hours were spent on 1412 packages in the archive (this is very little, but all that we could). Even if you consider that not all of this is an applications, a lot still needs attention. Open software culture is not typical on Amiga and this has only begun to change in recent years. Therefore, for most applications there is no source code, including those that are in the archive. Proprietary applications for m68k-Amiga written in assembler for AROS are generally lost. You can recompile two dozen applications that will not be enough for work. Before that you need to recompile the libraries, for example, SDL, SSL, etc. Here there are no people versed in cryptography, I myself passed only next to this topic. All have a SDL libraries of varying degrees of curvature (so no one uploads their versions). A somehow working ncurses relies on SDL, for example. It takes a lot of patches for this to work in an Amiga environment, which is low-level work. You can't just take and compile. Although as a first approximation with POSIX everything is in order, in practice, elementary things are missing in ABIv0 (!). For example, it's good if you know about this repository. Although this solution to only a few problems is incompatible with future AROS. And after that we say that we need to recompile everything, but the stable version of ABIv1 is being postponed for now.

There aren't any AROS-specific applications written in Assember, since everything that currently runs on x86-Aros is written with a high-level language and then compiled for the x86 architecture. In a ideal world, coder would just keep the sources, place them on a x64-AROS installation and rebuild for it. Real world, however, is far from being ideal. Many legacy applications were compiled when ABIv0 AROS was different than today, building them today would require a review of the code first. The step from 32 to 64 bit architecture, moreover if we keep under consideration the SMP-enabled version, would also require more attention. I perfectly understand this is not something the casual contributor would like to face. But when I say "no software" I don't mean that everything must be recompiled, but at least the bare minimum AROS programs we have: ZuneTools, MPlayer, OWB, AmiStart, all the whole Hollywood plugins and all the libraries they need to work. But if we hadn't even a working toolchain under x64, a working gcc devkit, how can we help ourselves?

[aGGreSSor]

There aren't any AROS-specific applications written in Assember, since everything that currently runs on x86-Aros is written with a high-level language and then compiled for the x86 architecture.

I correct my english text all the time and may be misunderstood. I wrote about applications for the Amiga, which are not and will not be on AROS, but which make the Amiga attractive. These are trackers, bitmap font editors, editors of music samples in IFF, and even damn Personal Paint (which i love), CEd, GEd, ProgED, etc. Answers to Today's Challenges: which text editors support markdown highlighting, for example? MUI-Vim? I have it constantly falling)))

ZuneTools

Maybe it can be moved quickly

MPlayer

Welcome to AHI (Alsa in original) and other libraries (lame, freetype, etc). Many patches.

OWB

Welcome to AmiSSL

AmiStart

No source?

all the whole Hollywood plugins and all the libraries they need to work

Need to financially interest the authors of Hollywood. Is it possible? No.

[Samurai_Crow]

ABI v0 was just placeholder code.  It didn't observe the DRY concept of "don't repeat yourself" by offering true pure residentable executables and libraries.  Without that you don't get the full performance that AmigaOS gave in the past.  Caches only have to latch on to one copy of things when the DRY concept is observed.
 Without the full feature set of AmigaOS, what point is there in having another OS?

Now getting on with the subject of whether I'm testing ABI v1, the answer to that is also no.  I think the boat has left the harbor for Amiga-like OS's in general because the need for SMP is vital in the age of multicore CPUs.  Is there room for yet another x86 OS when Haiku is better on all fronts other than GPU support?

Regarding Hollywood, if you need to run hosted on top of Linux just to get sufficient drivers to run AROS, why not just run the Linux version of Hollywood?

With AROS SMP, ABI v1 had potential but presently AROS, along with all the rest of the next-gen Amiga platforms, is only useful for nostalgia trips.  If it weren't for AmigaOS 4 and MorphOS stealing the thunder of AROS, there might have been a chance.  As it stands there is no chance for any of them.

I still crank up my old Amigas every now and then or FS-UAE sometimes but those days are becoming less and less frequent.  My MorphOS-based Mac Mini is occasionally started for running Hollywood on and testing. My MicroA1 is started as an idle curiosity sometimes.  AROS is started even less frequently yet.

Why are we sticking with our Amigas in this day and age but to stick it to Microsoft, Apple and the like?  They are almost ready to shift to Linux once WebAssembly catches on as a standard.  Once that happens all compatibility arguments will fall flat and capability will once again shine.  Aros lacks capability compared to the likes of Haiku, as I mentioned earlier, and WebAssembly requires a largely POSIX compliant ecosystem to implement.

The opportunity for small, streamlined OS's to shine is coming soon and sadly, we aren't ready for it.

[nikos]

Samurai_Crow: This is not about taking over the world. It is a powerful, cheap alternative to other NG Amiga Operative systems.
People have fun with C-64, Atari, Amiga classic and there is a lot of fun that can had with AROS. It is a hobby you can learn a lot from. I would not know anything near as much about computing if it was not for AROS.  I enjoy using it and enjoy any progress made to it. There is a lot of software where AROS do just as well as any other Operative system.  It is not that you need SMP, and the worlds best computer to run a calculator It is just to put things into perspective. My laptop is still core2duo using Peppermint 64-bit, no plan to upgrade. I play games mostly on C-64 and Amiga, that is where most of my favorite games are. I have a powerful Intel computer with Nvidia where I can play Dirt Rally II etc. but I do not use it much.

[Samurai_Crow]

@nikos

I ran Debian on my Core 2 Duo based Mac Mini for years using the LXDE desktop similar to your Peppermint Linux laptop.  I've since cross-graded to a G4 PPC Mac Mini to run MorphOS.  It runs great for a single threaded OS, but I'd never think of running a single-threaded OS on my second- or third-gen i7.  It would leave too many threads idle.  That's why I don't advocate running Aros on modern hardware at all.  If Aros SMP took off, it might have been different.

[nikos]

@nikos

I ran Debian on my Core 2 Duo based Mac Mini for years using the LXDE desktop similar to your Peppermint Linux laptop.  I've since cross-graded to a G4 PPC Mac Mini to run MorphOS.  It runs great for a single threaded OS, but I'd never think of running a single-threaded OS on my second- or third-gen i7.  It would leave too many threads idle.  That's why I don't advocate running Aros on modern hardware at all.  If Aros SMP took off, it might have been different.

Yes, I understand that. Running AROS native require some old supported hardware. I would never run AROS native in VESA mode, no matter how many cores I had supported. For new computers you can run hosted, but still no accelerated GFX, so it is a dead end for me. For development it can be nice. 

[OlafS3]

Hi,

I rarely commented here (mostly active on 68k related platforms) 

generally I would not be that pessimistic... Aros is used today by many users on 68k (V2 and V4) perhaps there is a chance to use that for the "bigger platforms". I remember when all sort of people explained me why aros and alsways will be a waste of time, nobody will ever be interested in and nobody will ever use  it as main platform. And that is exactly happening now. It happens because it works (on 68k) as a good replacement for 3.1, it behaves like 3.1 and much of the old software also works. Of course that cannot easily transferred to ABI1 (I would say with limited resources you should concentrate on one platform) but from a user point of view it should be as easy as possible and as similar to what people expect (amiga) as possible. Easy means to me there must be real reference hardware where it is possible even for unexperienced users to install and configure the OS. The hardware must be fully supported out of the box. Additionally Amiga users expect something amiga-like and want to run the old software so you need integrated 68k emulation. AROS should look & feel like Amiga did. Alternative you can try to attract other users but that is more difficult in my view.

So in short... if possible not separate platforms (ABI 0 and 1) but only one to get new developments but also testing as quick as possible and few selected modern hardware that is full supported (also f.e. RPi on ARM)

[braincure]

Olaf. The way of mind is one. I have expressed this too. Actually, what needs to be done for now is to prepare an Aros equipment list. An extremely common computer such as intel or amd, nvdia rtx 2060, realtek audio, intel wireless and bluetooth, and almost every user commonly owns it. and starting from here and showing itself in the real world should be the goal of Aros.

[nikos]

We already have this.

ACER aspire one 110 or Zg5 (same model).

For something more powerful I'm using DELL latitiude D520 laptop. What is needed is to change the wi-fi chip. It should for most be quite easy.
This laptop came in 3 different versions. I have changed the CPU on the 2 DELL laptops I have. I got the CPU on ebay for very little money. They all run Core2Duo 2GHz now. 1 or them I installed Peppermint the other AROS i386 ABI v.0.
Sure possible to have both operative systems on the same computer. Great laptops, with 2GHz core2duo they support 64-bit and both operative systems work fast and reliable.

[OlafS3]

I am aware of that... perhaps not enough promoted or not "sexy" enough for some users. Generally the way to go in my view. And as I wrote in another thread... Amiga users should feel at home when using aros, and 68k emulation and integration is critical

for me three things are critical:
easy to install and fully support hardware (most users are not able and interested in fiddling around with complex configurations)
Environment must look and behave like amiga
68k integration is critical... there must be no difference if you install and use X86 programs/games or 68k, the difference must be hidden

[nikos]

I am aware of that... perhaps not enough promoted or not "sexy" enough for some users. Generally the way to go in my view. And as I wrote in another thread... Amiga users should feel at home when using aros, and 68k emulation and integration is critical

for me three things are critical:
easy to install and fully support hardware (most users are not able and interested in fiddling around with complex configurations)
Environment must look and behave like amiga
68k integration is critical... there must be no difference if you install and use X86 programs/games or 68k, the difference must be hidden

Yes, sure a more modern solution would be preferd. Something where you could install at least one more modern operative system. My DELL laptop still kind of suits that bill, but there are some fiddling that most might not want.
Remember this been tried by "ARES" hardware and Stevens "IMica"
I'm sure this could be possible, but it would require a lot of promotion and software worked on.
It is also what I see as the right path to really kickstart AROS.
 

[cdimauro]

68k integration is critical... there must be no difference if you install and use X86 programs/games or 68k, the difference must be hidden

This isn't possible on architectures which aren't 32-bit AND big-endian (e.g.: PowerPC, or ARM32 running in big-endian mode).

[OlafS3]

it is not possible in a "petunia!" way but at least UAE integration should be as tight and transparent as possible

[nikos]

68k integration is critical... there must be no difference if you install and use X86 programs/games or 68k, the difference must be hidden

This isn't possible on architectures which aren't 32-bit AND big-endian (e.g.: PowerPC, or ARM32 running in big-endian mode).

You are right. It will never be transparent but it can get way better than it is now. A ready set up solution where partitions and the system is ready to run 68k software. I did that with AspireOS but the emulator is not on level with WinUAE and that is what it should, must be.