AROS Exec

General => General Chat => Topic started by: miker1264 on August 02, 2019, 11:58:44 PM

Title: What Could AROS Become?
Post by: miker1264 on August 02, 2019, 11:58:44 PM
Now that AROS is both native and hosted and supporting more hardware such as ARM and Pi what are it's strengths and weaknesses going forward?

Amiga 68k compatibility seems absolutely necessary in my opinion. Also having a fast and stable 68k version for 68k hardware such as Vampire accelerators is also important.

What do you think?
Title: Re: What Could AROS Become?
Post by: nikos on August 03, 2019, 02:34:27 AM
There been different goal and opinions about how AROS could be run.
Some tougt that AROS could be like Linux, supporting all kind of hardware.
That would be fantastic but would require a much bigger team of developers than
we ever had.
There was a dream that AROS could be a NG Amiga platform with all kind of software
running like on Windows or Linux but that is by far never going to happen. I'm not going
to explain why but there are to many limitations in the OS.

The way I see a future for AROS is as a hosted platform, running in a virtual player or on dedicated hardware like the Raspberry Pi.
It already does with AROS i386 version but we need the AROS 64-bit to the same level. It is what people use these days.

The Pi is great as it still in a state where AROS could fit in as an alternative OS.
On top of that it is very affordable. I seen some projects where it is even put into a laptop if that is what you want. The possibilities with Pi is endless. It has not stopped with Pi v.4 either. Who knows what Pi will bring to the future.

To sum it up.

Ask o1i how much he wants to port latest WinUAE to AROS. He already came quite far with a port of WinUAE for AROS 64 but that is broken now with some changes to the OS.
I know he is busy but if we offer and collect some good money it should be a motivator. Everything has a price.

Michael Schulz is working hard on the Pi port. It would supprise me if we will not see AROS running on Pi 4 within this year.
Kalamatee works on AROS on all aspects. If AROS 64 gets the love it needs both me and Paolo can bring our distributions to 64-bit. Personaly I'm most interested to do AspireOS 64 if we get some updated Amiga 68k emulator.   
 
AROS 64 distribution running on real hardware, hosted or WM with AROS 68k drives ready set to the standards of WinUAE 68k emulation could become quite popular. Other than running 68k classic stuff you would have a future proof platform.
That platform should focus on gaming, emulation, tracking and possibly some youtube player and media player. Fun computing as always Amiga has been.

Maybe even bring in some people from the demo scene to do a little production for AROS Pi.
Could be delivered to the wild compo at some party.
Title: Re: What Could AROS Become?
Post by: trekiej on August 03, 2019, 02:51:48 AM
Has Aros been a part of Google Summer of Code.
I really need to look at GSoC first before asking this.
edit:

Aros 64 SMP on RPi-4 would be awesome.(Native)
Title: Re: What Could AROS Become?
Post by: hth313 on August 03, 2019, 05:33:15 AM
Different people will have different opinions and goals. I think it is very important to have reference hardware that can be obtained easily.

I currently use AROS on 68k, but that is old hardware and it is not very fast. There are both good and bad sides about it.

Apart from that I would love to see AROS on 64-bit x86. But as I said, I think it is important to be able to either order a ready to use machine or use easily obtainable (from a shop) components.

As a reference platform I find the RPi very interesting for obvious reasons. It is cheap, easy to find should provide decent performance for many uses. Many people should be open for experimentation here and may not expect that it has to be a full blown modern desktop system.

In some way I see AROS as a platform much like an embedded system or set-top box. Something I can program to do certain specific duties, like controlling aspects of a model railroad or perhaps a slotcar race track system.

It can also be interesting to see how far an Amiga style operating system can be taken, but there are just perhaps too few developers to really see that become a reality.
Title: Re: What Could AROS Become?
Post by: o1i on August 03, 2019, 11:26:51 AM
Ask o1i how much he wants to port latest WinUAE to AROS. He already came quite far with a port of WinUAE for AROS 64 but that is broken now with some changes to the OS.
I know he is busy but if we offer and collect some good money it should be a motivator. Everything has a price.

Not me ;). Money can't buy me in this case, sorry. Time is the limiting constraint and you can't give me that.

Btw, I had not the impression many people were interested in the WinUAE port. I got exactly no feedback at all.

Maybe I'll continue with it in the future, but most likely not this year anymore.
Title: Re: What Could AROS Become?
Post by: miker1264 on August 03, 2019, 05:23:10 PM
Nikos

"The way I see a future for AROS is as a hosted platform, running in a virtual player or on dedicated hardware like the Raspberry Pi.
It already does with AROS i386 version but we need the AROS 64-bit to the same level. It is what people use these days.

The Pi is great as it still in a state where AROS could fit in as an alternative OS.
On top of that it is very affordable. I seen some projects where it is even put into a laptop if that is what you want. The possibilities with Pi is endless. It has not stopped with Pi v.4 either. Who knows what Pi will bring to the future.

To sum it up.

Ask o1i how much he wants to port latest WinUAE to AROS. He already came quite far with a port of WinUAE for AROS 64 but that is broken now with some changes to the OS.
I know he is busy..."

I agree with many things. I believe the way forward for AROS is different distros mostly Hosted on the Big Three and native on Pi or ARM. The 64bit version is necessary.

The Big Three as far as Hosted are Linux, Windows and Mac OS X. Virtual Machines such as VirtualBox is also useful. AROS native also has its uses. But mostly Hosted with Host Bridge offers exciting possibilities. We don't really need to write or port large elaborate programs like GIMP, Photoshop, Office, Blender or even a Web Browser. Host Bridge according to Paolone allows us to use the Host Programs. It's the best of both.

However, there is one path forward that's very important for AROS. The 68k version(s) should be made to be fast and stable to run on 68k hardware such as Vampire Accelerators.

As for developers having time to develop...its like when your wife has your first baby. Suddenly Free Time becomes a precious commodity. It's the same with programming.
Title: Re: What Could AROS Become?
Post by: miker1264 on August 03, 2019, 05:31:47 PM
Different people will have different opinions and goals. I think it is very important to have reference hardware that can be obtained easily.

I currently use AROS on 68k, but that is old hardware and it is not very fast. There are both good and bad sides about it.

Apart from that I would love to see AROS on 64-bit x86. But as I said, I think it is important to be able to either order a ready to use machine or use easily obtainable (from a shop) components.

As a reference platform I find the RPi very interesting for obvious reasons. It is cheap, easy to find should provide decent performance for many uses. Many people should be open for experimentation here and may not expect that it has to be a full blown modern desktop system.

In some way I see AROS as a platform much like an embedded system or set-top box. Something I can program to do certain specific duties, like controlling aspects of a model railroad or perhaps a slotcar race track system.

It can also be interesting to see how far an Amiga style operating system can be taken, but there are just perhaps too few developers to really see that become a reality.

I agree with that. It would be good to have a reference system, a baseline configuration that we know will work. Raspberry Pi and ARM might offer us some good opportunities for that.

Another important point is that AROS needs a cohesive and coherent set of development tools. A nice integrated designer for laying out MUI/Zune user interfaces, reference materials and lots of programming docs. Making it easier to program for AROS might attract more programmers. Just my thought.
Title: Re: What Could AROS Become?
Post by: miker1264 on August 03, 2019, 05:39:39 PM
Ask o1i how much he wants to port latest WinUAE to AROS. He already came quite far with a port of WinUAE for AROS 64 but that is broken now with some changes to the OS.
I know he is busy but if we offer and collect some good money it should be a motivator. Everything has a price.

Not me ;). Money can't buy me in this case, sorry. Time is the limiting constraint and you can't give me that.

Btw, I had not the impression many people were interested in the WinUAE port. I got exactly no feedback at all.

Maybe I'll continue with it in the future, but most likely not this year anymore.

I understand you have priorities. We all do to some degree. It's difficult to divide you time for different projects and still provide the same level of attention that each deserves.

Of the versions of UAE available I believe that WinUAE is the most advaced and most useful. I use WinUAE on a regular basis. A version of UAE for AROS based on WinUAE would be very useful. I'd like to try it when it's finished. Thanks for your efforts! Your time is appreciated.

Mike R.
Title: Re: What Could AROS Become?
Post by: salvatore on August 03, 2019, 06:10:44 PM
I use janus-uae 1.4 to make audio editig, and I could use other software without any problem, but I enjoy with aros and its resources (in fact I record my audio tracks with quickrecord), I do not know how much is worth the porting of this software, already version 1.4 is quite functional light and complete.
Title: Re: What Could AROS Become?
Post by: salvatore on August 03, 2019, 06:18:13 PM
I would think of a porting of janus 1.4 for x86_64 without overloading o1i of work, it also works well
Title: Re: What Could AROS Become?
Post by: miker1264 on August 03, 2019, 06:45:45 PM
I would think of a porting of janus 1.4 for x86_64 without overloading o1i of work, it also works well

While running Icaros Hosted on Windows if you have a shared folder with your 68k apps just use Host Bridge to launch WinUAE. As for Linux or Mac OS they would be limited to other versions of UAE to run the 68k apps.

In early versions of Windows it was possible to run 16bit apps in the console. Mac OS had a similar "sandbox" approach to legacy apps. Perhaps "seamless" integration of 68k apps is a worthy goal? When launching an app AROS detects whether it's Native or 68k. If it's a legacy 68k executable the app runs in the sandbox but behaves like a Native Program. The 68k sandbox in this case is a customized version of UAE integrated into AROS itself.
Title: Re: What Could AROS Become?
Post by: nikos on August 03, 2019, 06:59:55 PM
I would think of a porting of janus 1.4 for x86_64 without overloading o1i of work, it also works well

Janus 1.4 is ok. It just happend a lot regarding Amiga 68k emulation after Toni took over. What is very nice with Janus 1.4 is that it is very stable. For sure one of the most stable software on AROS. I have no idea if Janus 1.4 is ready for 64-bit?
The beta of Janus UAE2 is there for someone to compile. It is for sure ready for 64-bit.
Maybe o1i could guide someone how to do it?
 
Title: Re: What Could AROS Become?
Post by: salvatore on August 04, 2019, 12:14:59 AM
you need to make a list of applications to bring to x86_64, I personally, if not already started, would abandon janus v2
Title: Re: What Could AROS Become?
Post by: amigamia on August 04, 2019, 07:10:29 PM
I would think of a porting of janus 1.4 for x86_64 without overloading o1i of work, it also works well

While running Icaros Hosted on Windows if you have a shared folder with your 68k apps just use Host Bridge to launch WinUAE. As for Linux or Mac OS they would be limited to other versions of UAE to run the 68k apps.

In early versions of Windows it was possible to run 16bit apps in the console. Mac OS had a similar "sandbox" approach to legacy apps. Perhaps "seamless" integration of 68k apps is a worthy goal? When launching an app AROS detects whether it's Native or 68k. If it's a legacy 68k executable the app runs in the sandbox but behaves like a Native Program. The 68k sandbox in this case is a customized version of UAE integrated into AROS itself.

Wasn't this already implemented by Paolone on Icaros with Amibridge?

https://vmwaros.blogspot.com/2012/06/next-step-free-68k-integration.html (https://vmwaros.blogspot.com/2012/06/next-step-free-68k-integration.html)
Title: Re: What Could AROS Become?
Post by: Samurai_Crow on August 04, 2019, 11:10:32 PM
WinUAE is not as portable as FS-UAE and FS-UAE supports x86_64 already on Linux and Mac.  The only limitations are that the Windows version doesn't support the JIT.  Has anyone tried to port FS-UAE to AROS?
Title: Re: What Could AROS Become?
Post by: nikos on August 05, 2019, 01:58:49 AM
Ok, good to know it is more easy to port. Never been a version for AROS but JIT must be enabled.
If anyone is up to the task we could try to collect the money wanted to the port it.
Title: Re: What Could AROS Become?
Post by: o1i on August 05, 2019, 02:20:52 PM
Just some summary:

Porting of Janus-UAE v1.4 to 64 bit makes *no* sense. It is based on *very* old sources, which have never been compiled for 64bit anywhere at all. The v1.4 JIT is 32bit only, you would have to rewrite it from scratch for 64bit.

WinUAE *is* quite portable. In fact fs-uae is a straight port of WinUAE just with another GUI (more or less).

Janus-UAE v2 is a port of WinUAE including the GUI and working on 64bit.

There is no "shortcut" like just port v1.4 to 64bit.
Title: Re: What Could AROS Become?
Post by: aha on August 05, 2019, 03:37:52 PM
@o1i

I still hope that you can complete Janus-UAE v2 for 64Bit as a port from the WinUAE sources. Now Paolone is working on Icaros64 too and Janus would be an important part for this distribution imho. For the test of Janus-UAE v2 I had unfortunately no time the last couple of months because of my real life...
Title: Re: What Could AROS Become?
Post by: x-vision on August 07, 2019, 09:53:17 PM
There been different goal and opinions about how AROS could be run.
Some tougt that AROS could be like Linux, supporting all kind of hardware.
That would be fantastic but would require a much bigger team of developers than
we ever had.
There was a dream that AROS could be a NG Amiga platform with all kind of software
running like on Windows or Linux but that is by far never going to happen. I'm not going
to explain why but there are to many limitations in the OS.


If that comes to be true, I'm in kind of shock now. I know Aros is limited because the lack of resources, and missing modern features like multi user and memory protection (although I think there is some basic implementation), but I thought those would be addressable in the future, and that the basics and structure of the OS would be good enough to survive and improve for many years, because that was one of the goals: A modern re-implementation of Amiga OS.

If that's not true, I would be very dissapointed. What would be the point for Aros then?
Title: Re: What Could AROS Become?
Post by: miker1264 on August 07, 2019, 10:33:33 PM
There been different goal and opinions about how AROS could be run.
Some tougt that AROS could be like Linux, supporting all kind of hardware.
That would be fantastic but would require a much bigger team of developers than
we ever had.
There was a dream that AROS could be a NG Amiga platform with all kind of software
running like on Windows or Linux but that is by far never going to happen. I'm not going
to explain why but there are to many limitations in the OS.


If that comes to be true, I'm in kind of shock now. I know Aros is limited because the lack of resources, and missing modern features like multi user and memory protection (although I think there is some basic implementation), but I thought those would be addressable in the future, and that the basics and structure of the OS would be good enough to survive and improve for many years, because that was one of the goals: A modern re-implementation of Amiga OS.

If that's not true, I would be very dissapointed. What would be the point for Aros then?

It is my hope too that AROS contnues to advance and grow to become more modernized such as going to 64bit and support for multiple processors as well as drivers for usb 3.0/3.1 and improved graphics. But in order for all that to happen AROS must attract more developers.

Having an AROS IDE and a layout designer for quickly making user interfaces for MUI/Zune is important as well as more cohesive programming documentation. Making it easier to program may attract more programmers. Just my thought on the matter.
Title: Re: What Could AROS Become?
Post by: salvatore on August 08, 2019, 05:03:23 AM
Exactly Miker
Title: Re: What Could AROS Become?
Post by: o1i on August 08, 2019, 08:19:50 AM
There are not so many limitations in the OS design. It most likely will never get any secure runtime multiuser abilities and memory protection will never reach a level considered to be safe enough compared to other systems. It will always have one address space for all applications, but this limitation should be no problem anymore with 64bit pointers, as the 16 Exabytes address range should give you enough space to enlarge stacks for example. Currently not even started, but possible.

So there are possibilities to reach a certain level and if some hero steps up (which happens from time to time) speed can increase. Currently progress is slow, so the gap to modern systems doesn't get much smaller.

But for a developer this might be the challenge, push the limits further and further. That's what the Amiga always was, back then maybe with demos etc. Now there is an additional challenge for OS developers ;-)
Title: Re: What Could AROS Become?
Post by: dizzy on August 08, 2019, 09:27:43 AM
There been different goal and opinions about how AROS could be run.
Some tougt that AROS could be like Linux, supporting all kind of hardware.
That would be fantastic but would require a much bigger team of developers than
we ever had.
There was a dream that AROS could be a NG Amiga platform with all kind of software
running like on Windows or Linux but that is by far never going to happen. I'm not going
to explain why but there are to many limitations in the OS.


If that comes to be true, I'm in kind of shock now. I know Aros is limited because the lack of resources, and missing modern features like multi user and memory protection (although I think there is some basic implementation), but I thought those would be addressable in the future, and that the basics and structure of the OS would be good enough to survive and improve for many years, because that was one of the goals: A modern re-implementation of Amiga OS.

If that's not true, I would be very dissapointed. What would be the point for Aros then?

It is my hope too that AROS contnues to advance and grow to become more modernized such as going to 64bit and support for multiple processors as well as drivers for usb 3.0/3.1 and improved graphics. But in order for all that to happen AROS must attract more developers.

Having an AROS IDE and a layout designer for quickly making user interfaces for MUI/Zune is important as well as more cohesive programming documentation. Making it easier to program may attract more programmers. Just my thought on the matter.

The USB3.0 is almost there, it still needs a hubss.class (now merely a clone of hub.class) and memory allocator for the pcixhci.device (and transfer codes for it and a USB3.0 mechanism for the hubss.class to know the device/port state) USB3.0 provides better mechanism to do isochronous transfers so we should get that too at the same time the USB3.0 code shows some signs of live. Now the hubss.class binds with the USB3.0 driver, or at least it used to.

There are a lot of Forbid/Permit pairs to protect the Poseidon stack so it may hinder the performance a bit, a thorough cleanup of the USB code is also needed.

Transfer codes for a USB driver is nothing special, but we need some USB3.0 love for the driver(s). There are some bypasses in the Poseidon code, one for instance that the driver needs to tell to the stack what USB version it is used for, it's abit redundant as everything should be looked up from the roothub and its structures... It's all there.
Title: Re: What Could AROS Become?
Post by: salvatore on August 08, 2019, 06:43:53 PM
@o1i

but in my opinion there are many things in winuae that are not needed at the moment, all that part related to the emulation of various devices that are needed?
Title: Re: What Could AROS Become?
Post by: paolone on August 13, 2019, 02:10:33 PM
Janus UAE 1.4 is very old and it can't even run recent builds of AROS M68K. I am sticking to a very old 68K system base on 32bit AmiBridge because of this, and there aren't chances I will bug Oliver anymore to help me on this matter. I'd rather prefer him spending his own free time on Janus V2 instead: we must look at the future, not at the past. From time to time, Oliver is also helping me with porting/compiling software on both 32 and 64bit architectures and I shall be already happy with this.


Title: Re: What Could AROS Become?
Post by: salvatore on August 13, 2019, 02:30:08 PM
i understand, i use janus on abiv0 for one software for editing audio "samplemanager" and is essentially to have become janus for x86_64 i think really, attend next noticed for icaros good work

hi