AROS Discussion [moved from "AmigaOS 3.1.4 withdrawn from market after Amiga" thread]

[cloverskull]

Pie-in-the-sky:

AROS is interesting, but execution is tough because it seems like the entire thing was built from scratch, and (IMO) it just _feels_ wrong.

What would be fun, from my perspective, would be for every part of the amiga OS (rom components, workbench, etc) to be by opensource components piecemeal as we achieve parity. The only way to eat an elephant is one bite at a time.

Anyway I don't really think this would ever happen, but it would be great to reduce dependency on commercial organizations that can't find a way to work in a fashion that benefits the community.

[wawa]

Originally Posted by cloverskull

_feels_ wrong.

why? because the default skin is ugly? disable it, then it looks like 3.1.
because the icon theme is ugly? replace it.
because it runs on x86? run it on amiga or uae and use amiga software with it.

What would be fun, from my perspective, would be for every part of the amiga OS (rom components, workbench, etc) to be by opensource components piecemeal as we achieve parity. The only way to eat an elephant is one bite at a time.

people are already doing it piece by piece around here, and what does it lead to?
do you really think developing a complex system is better distributed among bedroom coders with no version control, no common repo, no build system and without possibility to check components against each other with every single change, because everybody cooks their own soup?

Anyway I don't really think this would ever happen, but it would be great to reduce dependency on commercial organizations that can't find a way to work in a fashion that benefits the community.

exactly. thats what aros is meant for..

[cloverskull]

I guess you don't understand what "pie-in-the-sky" is.

[Warty]

I haven't been doing anything with Amigas for 25 years or so (except playing with UAE from time to time), until just a few months ago. I kind of knew about Cloanto, Hyperion, OS3,9, OS4, AROS, etc., but haven't paid any real attention. I also don't know the folks on this board (except a few that have helped me in the coders section), so I'm a little hesitant to post, but sometimes it can be helpful to be an outsider.

This seems like an unfortunate situation.

Without having ever used AROS (except a couple months ago when FS UAE started up REALLY slowly the first time I installed it because I hadn't fed it a KS ROM), this is maybe a bit premature, but...

On the one hand, we appear to have an "almost-there" open source alternative, with some hard-working folks trying to do the impossible.

On the other hand, we have some legally-entangled parties that have been fighting over the carcass for quite a few years.

It's not like Amiga (technology) has any real commercial value at this point... This is a hobbyist situation. Maybe it's time to acknowledge that once and for all, set up a foundation for AROS (maybe one exists, but I mean a funded one with some specific goals), and make AROS be what we want it to be? (And that doesn't have to be one thing, we could have specific goals for legacy 68k improvements, 'modern' hardware support, etc., not necessarily opposed to each other, all working under the same umbrella.

Like I said: I don't know anything about AROS, so I have no idea if the people working on AROS would even be open to that, and if they would be open to the idea of optimizing code so that it could run fast on 68k. (Even if that was a 5 year effort, with other stuff going on at the same time). Anybody in a hurry here?

And yeah: I'm willing to put up some money, and I bet a lot of people would be. Look at how much money kickstarters raise! I know very few local Amiga people, and I know TWO that ordered the checkmate 1500 (part deux?).

[ExiE]

Originally Posted by wawa

why? because the default skin is ugly? disable it, then it looks like 3.1.
because the icon theme is ugly? replace it.
because it runs on x86? run it on amiga or uae and use amiga software with it.

So we know really well why is not AROS inviting enough.
Your suggested solutions are wrong from bigger point of view.

We should think like Haiku project, not linux.

If you look for "Amiga AROS" on google, first result is http://aros.sourceforge.net/ it looks official but there is some f* redirect to empty page http://ww5.arixfoundation.com/

Download link is hard to find and offers so many options that even I don't know which one is the right one (ABIv0 or 1, etc)

Default skin and icons should look attractive enough and be distinguishable. Classic OS3.1 skin is too old fashioned and default skin and icons are really ugly...

[Dunny]

Originally Posted by ExiE

Default skin and icons should look attractive enough and be distinguishable. Classic OS3.1 skin is too old fashioned and default skin and icons are really ugly...

You hit the nail right on the head there and I can only nod in agreement. AROS should be MagicWB through and through.

[Heiroglyph]

Maybe hold off on the look until Aros can draw the Workbench menu in under 3 seconds an 25MHz 040.

A 7MHz 68000 can do it instantly with AOS, but since Aros focuses on 4GHz PCs you end up with crap like watching a menu draw scanline by scanline for a few seconds on a fast 68k system.

Imagine what it could do on your beloved PCs if it could instantly draw a menu on a 7MHz 68000?

I want Aros to work, but some of it is pretty horrible and 68k seems to be barely on the development radar anymore, they'd rather add SMP to the PC port.

[Samurai_Crow]

Re:AROS graphics drivers
The graphics drivers for the Amiga chipsets need a ton of work. That's true. The chunky drivers also need some work but not as bad as planar.

As far as SMP being added, it seems that there are only three guys working on the AROS project other than Wawa and the RasPi 2 and 3 are targets of Dr. Michael Schultz. PC hardware is too expensive for AROS development anyway. PPC is even worse.

[wawa]

Originally Posted by Heiroglyph

Maybe hold off on the look until Aros can draw the Workbench menu in under 3 seconds an 25MHz 040.

are you talking about planar amiga screen? defaut aros skin renders translucent menus via alpha channel, which will never be fast on amiga chipset. you simply need to disable decoration. have you tried that? its done either in prefs/appearance or deleting theme.var file in env/sys.

here is a couple of videos of aros on vampire, which doesnt look exactly that slow:
[ Show youtube player ]
admittedly, its the rtg/saga output, but with decoration switched on in this case.

now. with aros-m68k people have developed a strange attitude. even if they are basically willing to help, they first make demands, that this or that needs to be improved first. as a person who was helping with aros-m68k since day one, im asking myself, what is value of such proposals? what would the people need to help with if everything they demand would have been delivered up front.

[wawa]

Originally Posted by Heiroglyph

I want Aros to work, but some of it is pretty horrible and 68k seems to be barely on the development radar anymore, they'd rather add SMP to the PC port.

smp has been added years ago. and m68k is still on a development radar, as far as i know and am concerned, as much as everything else. last version of ata device i have been testing had skeletons for improved native support of amiga addons such as fastata. other than that michal is currently developing big endian arm/pi target with a trancparent m68k emulation layer in outlook, such as is used on os4 or morphos. but people get dis encouraged if there is only bashing and no real interest otherwise.

[Heiroglyph]

Originally Posted by wawa

are you talking about planar amiga screen? defaut aros skin renders translucent menus via alpha channel, which will never be fast on amiga chipset. you simply need to disable decoration. have you tried that? its done either in prefs/appearance or deleting theme.var file in env/sys.

I didn't know to try that, I just built it, got it on one of my A4000 systems and tried AGA and RTG screen modes.

Maybe that shouldn't be the default on 68k builds?

here is a couple of videos of aros on vampire, which doesnt look exactly that slow:
[ Show youtube player ]
admittedly, its the rtg/saga output, but with decoration switched on in this case.

That is the fastest CPU currently possible plus RTG, so I'm not sure how to compare that. I've seen those playing videos that would never play on a stock system.

now. with aros-m68k people have developed a strange attitude. even if they are basically willing to help, they first make demands, that this or that needs to be improved first.

I don't think that running on the intended platform at an acceptable speed is an outrageous demand.

I've always felt that running on Amigas was an afterthought that has delayed development and made actual Amiga users not pay attention to the project. That's my basic reason for not being interested in it until finally giving up on any of the commercial OS versions based on the Commodore source code.

as a person who was helping with aros-m68k since day one, im asking myself, what is value of such proposals? what would the people need to help with if everything they demand would have been delivered up front.

It's too late now, but if they had written a kickstart replacement for 68k first before writing for a completely foreign platform, they would have had something usable much more quickly, possibly before most of the Amiga coders left. It wouldn't have looked so hopeless for so long that nobody wanted to sign up.

Instead, they ported the whole platform and supporting software to a super fast little endian system that couldn't have been more different and now try to shoehorn that onto Amigas. The PC can power through some really horrible code that would be immediately obvious on a slow CPU and it has no RAM and ROM size limitations that would have encouraged writing small code.

I'm ready to start giving AROS donations rather than lining another lawyer's pockets, but lets not pretend AROS is remotely usable on an average Amiga. It needs a lot of work.

[Samurai_Crow]

Originally Posted by Heiroglyph

I'm ready to start giving AROS donations rather than lining another lawyer's pockets, but lets not pretend AROS is remotely usable on an average Amiga. It needs a lot of work.

See http://power2people.org for bounty details.

[Heiroglyph]

Originally Posted by Samurai_Crow

Re:AROS graphics drivers
The graphics drivers for the Amiga chipsets need a ton of work. That's true. The chunky drivers also need some work but not as bad as planar.

Maybe things have improved since I tested, but they were barely functional on my system when I last tried and I haven't heard any reports of big improvements.

I'll gladly admit I'm wrong if that's the case.

As far as SMP being added, it seems that there are only three guys working on the AROS project other than Wawa and the RasPi 2 and 3 are targets of Dr. Michael Schultz.

And I'm immensely grateful for Wawa's continued work on 68k. Toni and Jason too back when they were making it somewhat work on 68k. Those are the only ones I know by name, but anyone else who gave attention to AROS on Amigas deserves to be thanked.

PC hardware is too expensive for AROS development anyway. PPC is even worse.

I can't agree there. decently fast PC hardware gets thrown away every day and the newest hardware isn't supported anyway.

The relatively cheap Amigas don't have a chance at running AROS as it stands, you need a $1000 in upgrades just to make it run. It needs to be developed with those in mind and we know it can work because we've seen it since the 80's.

For a PC OS that has to use UAE to run 68k code I'll just stick with Linux or Windows, I only care about AROS on real Amigas.

I'm not bashing, just trying to be realistic about what is missing for using it on Amigas so maybe the problems can be addressed.

[Romanujan]

My 0.02 EUR (0.03 EUR in fact).

1. AROS m68k. It has gained A LOT of users years ago, many of them still use it today, they talk about problems they have on the forums, etc. Really. The AROS I'm talking about is called AfA OS. Small problem: it didn't see any updates since years. The only m68k AROS updates go to the 'standalone' AROS which, performance wise, is just terrible - not everyone has 68060 + RTG, not everyone has the Vampire; 68030 + AGA or 50 MHz 68000 + ECS are very common configurations, and if AROS does not work on them with decent speed... well

I used AfA OS too - but when PeterK great icon.library emerged, and I couldn't use it with AfA without losing most of its features, and AfA were no longer updated, I have decided to go without AfA.

IMHO: you want m68k users? give us a new AfA! Not updating what people want to use, and updating what people don't want to use is... not the way to gain/keep users

2. AROS x86. What motherboard should I buy for the Ryzen system I'm going to build? It runs on UEFI, right? Will it work on my GTX 1070 with audio over HDMI? Won't it kill my SSD? Is it stable enough that it won't damage non-AROS partitions? If the answer to all these questions is: "hey, hey, hey, we don't have people, we don't have cash for the hardware, we can't..." - then just ditch the x86 native port completely. It's not really that useful anyway, if I need to use UAE to run just about any Amiga software... and it only drains the very limited developers time.

3. AROS on Raspberry Pi - this is a good idea, especially when transparent m68k emulation will be implemented. Hardware is cheap enough that a lot of people might buy it just for curiosity. Just make sure people get informed when there is something to test for the average user!

[kolla]

I have AROS on MiSTer, with scsi.device 43.45, pfs3aio and PeterK's icon.library and DOpus 5.91. It works quite well. For fun I also assigned in libs and classes directory of OS3.9 to the LIBS: so Reaction programs work well too.

[Heiroglyph]

Originally Posted by Romanujan

My 0.02 EUR (0.03 EUR in fact).

1. AROS m68k. It has gained A LOT of users years ago, many of them still use it today, they talk about problems they have on the forums, etc. Really. The AROS I'm talking about is called AfA OS. Small problem: it didn't see any updates since years. The only m68k AROS updates go to the 'standalone' AROS which, performance wise, is just terrible - not everyone has 68060 + RTG, not everyone has the Vampire; 68030 + AGA or 50 MHz 68000 + ECS are very common configurations, and if AROS does not work on them with decent speed... well

I used AfA OS too - but when PeterK great icon.library emerged, and I couldn't use it with AfA without losing most of its features, and AfA were no longer updated, I have decided to go without AfA.

...

I'm not going to quote all of your reply because I agree with almost all of it.

AfA patching into AOS doesn't free us from the old Commodore licensed code that is such a lawyer magnet, which is the reason for this thread.

Even with zero improvements to the 3.1 ROMs and floppies, having an open source 1:1 replacement for 68k Amigas would make these lawsuits irrelevant. That's what I'm interested in.

Improvements can come later, but you have to have a solid base to start from.

[Heiroglyph]

Originally Posted by kolla

I have AROS on MiSTer, with scsi.device 43.45, pfs3aio and PeterK's icon.library and DOpus 5.91. It works quite well. For fun I also assigned in libs and classes directory of OS3.9 to the LIBS: so Reaction programs work well too.

I'd like to know what parts you can use to make a usable system. I didn't realize it was compatible enough for that and that some required AROS libraries weren't the main problem.

That could point out where the code needs improvement.

[kolla]

Well, I have no clue about what you and other want or need, but AROS is there for anyone to fiddle with, just get started already.

[wawa]

Originally Posted by Heiroglyph

I didn't know to try that, I just built it, got it on one of my A4000 systems and tried AGA and RTG screen modes.

btw. keep in mind aros builds currently mc68000 binaries. as toni said debugging first, optimizing - later. but then if so opimizing may never occur.
one of the latest things kalamatee made me commit was adding x86 isa flags to the build system and indicated m68k should receive the same. it probably should be looked into soon.

Maybe that shouldn't be the default on 68k builds?

yes, but the developers argue that defaults should be in accordance in all platforms for better comparison. delivering a distribution on the other hand the autor can omit this one file, along with just few changes and maybe another icon set. at least something like that shouldnt prevent experienced coders to contribute.

That is the fastest CPU currently possible plus RTG, so I'm not sure how to compare that. I've seen those playing videos that would never play on a stock system.

its just an example. im testing arros on a stock system so to say, if you consider an a1200 or a4000 with some accel a stock system.

I don't think that running on the intended platform at an acceptable speed is an outrageous demand.

its not, for a user. we should definitely work towards that. but it is an area experienced 68k coders might exactly contribute.

I'm ready to start giving AROS donations rather than lining another lawyer's pockets, but lets not pretend AROS is remotely usable on an average Amiga. It needs a lot of work.

i wouldnt support bounties as long there arent a coder who is appointed to pick them up. it would be waste of time and money. but you can try to motivate the devs other ways, liker supporting michal schulz on patreon:
https://www.patreon.com/michal_schulz
or talking to kal, maybe he could consider to return, as he still was giving me bits and pieces of his code for a time. kal has no development system he needs, and unfortunatelly he has been bashed on because he outlined what was practicable to continue..
last but not least deadwood has made appearance again stating that he wants a vampire standalone. maybe he wants to do something with it. we will see.

[wawa]

Originally Posted by Romanujan

The AROS I'm talking about is called AfA OS.

i used afa for years and worked together with its autor, bernd.
he is posing here now and then.
but afa is another way of patching amiga os, imho ideally the os and kickstart should though be replaced.
but then i dont want to stop anyone to convince bernd to update afa.