Free AmigaOS for all

[welck]

Hi guys,
First time posting here but it's a post I also placed on the AmigaOS site. No reason not being well informed :-D I've been thinking; is there any incentive to make older Amiga OS's (like OS 3.5 and before)  freely available for download? Preferably for any platform? :inquisitive:

I can't see any problems for the industry; the AmigaOS would spread freely among users and software companies can still sell their Amiga programs.

What if we and/or Amiga companies alike could buy the rights from KMOS inc, have it compiled for all the platforms and have it freely distributed. Beacuse I can't see Amiga OS being commercially exploited (competing with Windblows) but definitly the great programs that it runs!

It wouldn't have the problems of an open source because the source would be in hands of a consortium of companies or users which could have agreements about standards and common development. Wouldn't that be sweet?

I wouldn't even see the problem in running native PC and Apple software too! Doom3 or connecting an IPod on your Amiga!

Anyone up for an incentive like this? Or is this only a daydream?  

[Ilwrath]

Well, AmigaOS itself is tied up in so many ownership disputes and original patent/licensing problems, it'll never be freely available.  

Even if by some chance you managed to get ahold of the 3.1/3.5 source tree.... Ask Hyperion what kind of condition it was in when they recieved it.   From what I understand, it was a complete mess.  It took them a couple months to just sort through it and get a semi-reasonable CVS together.  Tons of not-so-portable BCPL code, too.  

Basically, what I'm getting at here is that a free AmigaOS of any version level isn't going to happen.  Period.

Now, AROS....  Is basically exactly what you are wanting.  An open-source re-implementation of the AmigaDOS system on much more portable code base.  It's actually progressing quite well.  Check out the AROS MAX 0.2.2 listed in the news section here.  :-)

[bloodline]

ILwrath is right, the source code to AmigaOS up to 3.9 is pretty much 68k assember (so it can't be ported to other CPU's) and, from what I understand, in a very bad state.

Also, I belive, the source for the OS eariler than 3.1 has been lost.

The AROS sources are freely available though, and if it weren't for the ease of downloading a Kiskstart ROM from the net (:-(), they would have been used to make a free kickstart for Emulators etc...

[Karlos]

What's the current state of AROS on 68K / PPC ?

[bloodline]

Karlos wrote:
What's the current state of AROS on 68K / PPC ?

68k port works, but only on the Palm (and 68k linux hosted), as no one has ported AROS to the Amiga hardware :lol:

Hit a stumbling bock on the PPC, though as it seems that the varargs function has different behaviour on different platforms (passing parameters in registers or on the stack)... the best option would be to remove it from the AROS core, but apparently that would take a huge amount of work :-(

[welck]

AROS looked very interesting but those poor guys are still a bit understaffed :roll: So actually making the Kickstart ROM freely available (and perhaps open sourced) could solve these issues?

[SamuraiCrow]

@welck
Perhaps you don't understand.  AmigaOS 4.0 is currently being developed as a commercial piece of software for the PowerPC.  Unlike all the previous versions it has a hardware abstraction layer to make it portable to any platform that pays the licencing fee (for which they get to call their computer an Amiga).

[welck]

Yes that sounds pretty sweet but still; wouldn't it be oke to release the old versions for free on the internet then (for everyone to play with or something)? And how about the problems commercially competing with the other operating systems? :sealed:

[Karlos]

bloodline wrote:

Karlos wrote:
What's the current state of AROS on 68K / PPC ?

68k port works, but only on the Palm (and 68k linux hosted), as no one has ported AROS to the Amiga hardware :lol:

Hit a stumbling bock on the PPC, though as it seems that the varargs function has different behaviour on different platforms (passing parameters in registers or on the stack)... the best option would be to remove it from the AROS core, but apparently that would take a huge amount of work :-(

Mmm. Since it's an AmigaOS compatible project, why weren't TagLists used for variable length argument calls? Admittedly they use more stack but the idea works well enough.

Are there a lot of varargs functions? Other than for formatted IO and wrapping Taglist calls in a more C compliant way, are they essential?

[bloodline]

Karlos wrote:

bloodline wrote:

Karlos wrote:
What's the current state of AROS on 68K / PPC ?

68k port works, but only on the Palm (and 68k linux hosted), as no one has ported AROS to the Amiga hardware :lol:

Hit a stumbling bock on the PPC, though as it seems that the varargs function has different behaviour on different platforms (passing parameters in registers or on the stack)... the best option would be to remove it from the AROS core, but apparently that would take a huge amount of work :-(

Mmm. Since it's an AmigaOS compatible project, why weren't TagLists used for variable length argument calls? Admittedly they use more stack but the idea works well enough.

Are there a lot of varargs functions? Other than for formatted IO and wrapping Taglist calls in a more C compliant way, are they essential?

Um... I think Varargs were used to implement taglists :-D

I'm afraid I don't know much about them, I keep way from the complex bits, as I tend to break them (I have a histroy there :-)).

[Karlos]

Buh...

If you have an array of tags on the stack, terminated with TAG_DONE, 0, why do you need to do any dodgy varargs stuff anyway? It's no different in many ways to a zero terminated C string :-/

[bloodline]

Karlos wrote:
Buh...

If you have an array of tags on the stack, terminated with TAG_DONE, 0, why do you need to do any dodgy varargs stuff anyway? It's no different in many ways to a zero terminated C string :-/

Well this isn't my area, but the problem is due to not knowing if the parameters are on the Stack or in the registers... as on the x86 they are most likey passed on the stack, on the PPC they are more likely passed in registers (untill you run out, then they go on the stack). There are solutions, using a varargs macro or applying Fabio's patch to gcc to always pass parameters on the Stack... but for some reason these are not optimal solutions.

Anyway... how do you fancy writing some drivers for an Amiga classic port of AROS :-D

[Karlos]

Since a TagList is an array of struct TagItem, I'd be amazed if it was not passed by reference on any platform. Still, when you have so many registers as PPC I can see it might occur for short taglits.

As for the driver thing, no chance - I get only a few hours a week to work on my own stuff as it is and thanks to work I generally can't be bothered even then :-(

[bloodline]

Karlos wrote:
Since a TagList is an array of struct TagItem, I'd be amazed if it was not passed by reference on any platform. Still, when you have so many registers as PPC I can see it might occur for short taglits.

Um... OS coders are a strange and mystical bunch... They do things I'll never understand....

As for the driver thing, no chance - I get only a few hours a week to work on my own stuff as it is and thanks to work I generally can't be bothered even then :-(

Well, a good start would to make a loader that can bring the machine from power on to loading and executing Exec, without a Kisckstart being present... I'd be keen to have a go myself, but I can't figure out how the AROS build preocess works (it's been explained to me 1000's of times), and I'm even more in the dark when it comes to cross compiling... I like to stay in userland :-)

Surely you must have just such a loader lurking on your miggy HD somewhere? :-D

[Karlos]

Hmm. I'm not about to attempt the reverse engineering of the OS4 bootloader if that is what you are hinting at :-P

Even if I lacked the requisite scruples, long lists of PPC asm does my nut in :lol: