To AROS team

[hbarcellos]

Hello all. I was reading the "So what does the community want" post and faced ChuckT post there.

He wrote some lines that I personally agree, like this one:
"The magic behind the computer was that you could write your own programs and make your own graphics"

Some other points to form a conclusion:

1) Pretty much 90% of the amiga.org users would not buy one of the expensive solutions for running OS4 neither Morph OS.
2) AROS runs fine inside vmware, but... Installing it natively is a nightmare. Just tried on my quad core and mouse was not working...
3) MorphOS and AmigaOS4 can run more stable in part due to the narrow hardware database that they're supposed to find: Sam440p, Efika, Amiga PPC cards, etc...
4) Would probably take a LOOOOOOOOONG road to make AROS support most of the huge universe of different pc hardwares.

SO: why not just take a common x86 PC configuration from some years ago: Let's say Athlon Barton XP 3200+ with Radeon 9600, Sound Blaster Live+, nforce2 chipset and focus just on that?
We would be somehow back to "custom chips" that made Amiga so famous: AMD Barton being the new 68k, ATI RV360 being the new Denise and so on..

1) Enthusiastics, like most of us, would be able to assemble a 8000+ MIPS machine cheaply and still blazing fast!
2) Would drastically reduce development and mainly testing time for the development/testing team
3) Someone could come up with some custom cases (like the ones made to Minimig)
etc...
etc...
etc...

[Ranchu]

Most of the components you list are already well supported (AthlonXP, nforce2, and possibly SB Live+).  Just substitute an older Nvidia card for the ATI and you are set.

Of course you can't buy all this hardware new in the stores any more but there'a plenty of it around...

[fishy_fiz]

Actually Radeon 9600 is also supported in ATI driver (as are some newer ATI cards). I actually used to use one in my AROS box. SBLive is supported too. Coincidently its also the sound card I use for AROS. Seem to be a few of these "why dont they just support......." type threads lately in regards to AROS, which seems a bit unusual to me, as its not all that difficult to put together a fully supported system (athlon 64, gf5900,sblive, intel 100 pro nic in my dedicated AROS box, all supported in native). What I think could be useful is some sort of website that lists what hardware is supported to what degree. This way people could put together an AROS system without worry about things not being supported. I also think a bare minimal custom linux hosted version would be good (just enough to host AROS and have linux invisible to user (ala amiga forever/amithlon)) as it'd make a lot more hardware available to AROS users.

Just my 2 cents :-)

[dammy]

I also think a bare minimal custom linux hosted version would be good (just enough to host AROS and have linux invisible to user (ala amiga forever/amithlon)) as it'd make a lot more hardware available to AROS users.

Then come on over to #anubis channel for a visit.  

Dammy

[hbarcellos]

 Ok, but how much effort is being spent supporting all different x86 HW types?
 My point is narrow the supported hardware list to minimize effort trying to solve compatibility issues to focus on functionality (i.e. Binary Amiga compatibility).
 Looks to me that ppl prefer to set "boxes" apart from their x86 systems, which is more like a domestic utility this days.
 Tell me something: If you guys restricts even more the hardware, let's say:
 MB: Asus A7V880 - Chipset Via KT880
 VIDEO: Radeon 9600XT - RV360 Chip
 Sound: SB Live+ (EMU10K1 and FX8010 chips)
 Processor: AMD Athlon XP (Barton Core)

 Wouldn't that allow the OS and even some third part software like Demos to trick that hardware to the maximum, tweaking it, using non-documented features and etc?

[SamuraiCrow]

When the specific hardware model you specified has one component discontinued, the non-documented features will vanish.  This will leave you with incompatible software on all counts.

I'd be happy if AROS could even have hardware accelerated OpenGL or Mesa.  This is impossible with current licensing restrictions for the graphics chipset documentation because the graphics chipset manufacturers do their drivers in-house and don't support 3rd party operating systems.

[hbarcellos]

- "When the specific hardware model you specified has one component discontinued, the non-documented features will vanish."

When that happens, OS team could consider to launch a newer revised model: Core 2 Duo, Radeon HD, ...
 A-Box 500 would be replaced by A-Box 1200!

other comment about that:
MSX is discontinued. Non documented z80 features are all around.
ZX Spectrum, Amiga, C64 and etc...

" This is impossible with current licensing restrictions for the graphics chipset documentation because the graphics chipset manufacturers do their drivers in-house and don't support 3rd party operating systems"
 - It's being done by AROS, MorphOS and Amiga OS 4. And if you "at least" have a non moving target (fixed configuration A-Box 500) it should be easier to reverse engineer/debug stuff...

[Piru]

@SamuraiCrow

I'd be happy if AROS could even have hardware accelerated OpenGL or Mesa. This is impossible with current licensing restrictions for the graphics chipset documentation because the graphics chipset manufacturers do their drivers in-house and don't support 3rd party operating systems.

http://developer.amd.com/documentation/guides/Pages/default.aspx#open_gpu

[hbarcellos]

fishy_fiz wrote:
Actually Radeon 9600 is also supported in ATI driver (as are some newer ATI cards). I actually used to use one in my AROS box. SBLive is supported too. Coincidently its also the sound card I use for AROS. Seem to be a few of these "why dont they just support......." type threads lately in regards to AROS, which seems a bit unusual to me, as its not all that difficult to put together a fully supported system (athlon 64, gf5900,sblive, intel 100 pro nic in my dedicated AROS box, all supported in native). What I think could be useful is some sort of website that lists what hardware is supported to what degree. This way people could put together an AROS system without worry about things not being supported. I also think a bare minimal custom linux hosted version would be good (just enough to host AROS and have linux invisible to user (ala amiga forever/amithlon)) as it'd make a lot more hardware available to AROS users.

Just my 2 cents :-)

btw, I tried to run latest AROS Live CD on my personal "domestic utility" but my USB mouse was not functional. M$ Mouse and Keyboard plus Core 2 Quad on one of those ASRock Boards with dual AGPxPCI-X slots (Geforce 6800GT AGP)...

[uncharted]

It's so simple...

Target AROS to a VM (UAE) that can be hosted under any OS, so everyone can benfit from it.    For a 'native' solution simply host under a minimal Linux distro.

No more need for Amiga Inc.and thier 'IP'.
No more Driver headaches.
No more faffing about trying to reinvent the wheel.
Ridiculously high compatibility with the existing software library.


For that extra kick simply have x86 binaries a la Amithlon (which is apparently possible in UAE) for CPU critical tasks.

[hbarcellos]

uncharted wrote:
It's so simple...

Target AROS to a VM (UAE) that can be hosted under any OS, so everyone can benfit from it.    For a 'native' solution simply host under a minimal Linux distro.

No more need for Amiga Inc.and thier 'IP'.
No more Driver headaches.
No more faffing about trying to reinvent the wheel.
Ridiculously high compatibility with the existing software library.


For that extra kick simply have x86 binaries a la Amithlon (which is apparently possible in UAE) for CPU critical tasks.

Well, they're probably trying to do that with VMWAros but to me, that looks like WinUAE. And as I said, ppl apparently prefer to build their own "Amiga" boxes. And "native solution simply host under a minimal Linux distro" sounds like a huge waste of resources. AND, probably Anubis will be much better in that direction.

[cicero790]

Edit

[iconoclaST]

AROS is a step in the right direction...surely better than Amiga Inc's offering. I'd like to see AROS come completely off its training wheels (linux hosted)...I don't see why leaving it in, or making it more dependent on a VM would be a good thing.

Though 3.1 source compatibility seems to be the current emphasis of the development team, coming before maturing it enough to ride the bicycle itself. Which, is an understandable priority.

One thing I wonder though, out of curiosity...would anyone here consider running an OS that was very similar to the look and feel of AmigaOS, but threw compatibility with legacy apps completely out the window? (a la Mac OS switch to Mac OS X)

I got a crazy idea, and I've been trying to give myself some exercise in C coding anyways...though I'm not promising anything would become of it.

[dammy]

I'd like to see AROS come completely off its training wheels (linux hosted)

AROS has been running natively for years.

Dammy

[iconoclaST]

I'm aware of that, but you see less support under native mode than you do under Linux. I used to have a box dedicated to running AROS, but on my ThinkPad, it's a no-go.

My personal opinion is that (Well, I feel like tomatoes are going to be thrown at me for saying this) while AROS does well with focusing on 3.1 source compatibility, and while the original AmigaOS most definitely had its strengths, I couldn't really envision it as anything that's really ready for the 21st century. I would personally like to see something along the lines of the concepts of AmigaOS and intuition being injected into an existing *NIX-based OS distribution, and leaving behind backwards compatibility. It wouldn't be an /easy/ task, but I think think it would be, at the very least, doable and interesting. Not to mention that you could reap the benefits of thousands upon thousands of existing device drivers and such...

Take a *NIX distribution, develop a custom windowing system, re-work the filesystem layout, commands, and the feedback/syntax of commands to mimic the AmigaDOS shell, but while taking advantage of the heightened security and user privilege system, among other things...

The overall concensus is still, these days, the mainstream user just isn't ready for a *NIX-based OS that runs on common PC hardware. (Okay, I'll admit I'm shooting for the moon here, and not really taking myself too seriously), but something like say... "Intuix" or whatever it would be called...would at least be a good step closer...

A very unlikely proposal, but hey, one can always dream and try to get better at C in the process, can't they?