Welcome, Guest. Please login or register.

Author Topic: aros fork?  (Read 4324 times)

Description:

0 Members and 1 Guest are viewing this topic.

Offline Einstein

  • Sr. Member
  • ****
  • Join Date: Dec 2004
  • Posts: 402
    • Show only replies by Einstein
Re: aros fork?
« Reply #14 on: February 06, 2008, 01:23:58 PM »
Quote

Crumb wrote:

IMHO a fork is mandatory.

A new OS shouldn't have to deal with legacy problems caused by the old 3.x API.

IMHO there's no sense in calling the new effort AROS since it won't have anything related to the original effort. It won't be a v2.x version since it would be totally incompatible and probably won't reuse much code so it will be a total rewrite.

It would be in the same league as other kernels and OSes and probably could take advantage of code written for POSIX platforms.

Being "POSIX" compatible would be a good goal.


This sounded more sensible, gotta say your previous post left me  :crazy:
I have spoken !
 

Offline bloodline

  • Master Sock Abuser
  • Hero Member
  • *****
  • Join Date: Mar 2002
  • Posts: 12113
    • Show only replies by bloodline
    • http://www.troubled-mind.com
Re: aros fork?
« Reply #15 on: February 06, 2008, 02:23:11 PM »
As I see it "AROS2" as we now seem to be calling it, will still be aros but with the cruft removed... exec and all the exec style structures will still be there, but some stuff is going to need the change to support new technologies and these changes will introduce incompatibilities... A price I'm prepared to pay!

Offline itix

  • Hero Member
  • *****
  • Join Date: Oct 2002
  • Posts: 2380
    • Show only replies by itix
Re: aros fork?
« Reply #16 on: February 06, 2008, 02:52:20 PM »
Quote

but when it does (E-UAE) then that can be put on top of the new system


That is not different from EUAE on Linux or WinUAE on Windows.

Even more so if you break API compatibility you can not integrate UAE into OS more than you could integrate UAE into Windows.

Quote

I see, if one breaks backwards compatibilty at soure/binary level then it's "anything but amiga-like" (quoting you from above), yet, somehow magically, a different *premade* OS *is* !


If somebody wrote Windows clone which is not binary or source compatible with Windows, is it Windows at all? ;-)
My Amigas: A500, Mac Mini and PowerBook
 

Offline Einstein

  • Sr. Member
  • ****
  • Join Date: Dec 2004
  • Posts: 402
    • Show only replies by Einstein
Re: aros fork?
« Reply #17 on: February 06, 2008, 06:04:04 PM »
Quote

itix wrote:
Quote

but when it does (E-UAE) then that can be put on top of the new system


That is not different from EUAE on Linux or WinUAE on Windows.


Can you open WinUAE/AmigaOS windows as host (Windows) windows ? screens ? not even these ? how does it compare then ?

Quote
Even more so if you break API compatibility you can not integrate UAE into OS more than you could integrate UAE into Windows.


Depends what your definition of "integration" is, for me it's files, windows, screens. So, why could this not be on top of an API incompatible OS ? files are already integrated in WinUAE at least, sure it would be more code invloved than on a API compliant or -like system, but nevertheless, what would be the obstacle ?

Quote

I see, if one breaks backwards compatibilty at soure/binary level then it's "anything but amiga-like" (quoting you from above), yet, somehow magically, a different *premade* OS *is* !


If somebody wrote Windows clone which is not binary or source compatible with Windows, is it Windows at all? ;-)
[/quote]

If you had a blind cousin and if he got his vision back but suddennly and unfortunately became deaf would you not still regard him your cousin ? you see, you would (learn to) communicate differently with him now, in an "incompatible" way.
I have spoken !
 

Offline uncharted

  • Hero Member
  • *****
  • Join Date: Mar 2002
  • Posts: 1520
    • Show only replies by uncharted
Re: aros fork?
« Reply #18 on: February 06, 2008, 09:22:49 PM »
Quote

bloodline wrote:

now is a good time to decide what we want from AROS... the floor is open...


I have to disagree.  10 years ago would have been a good time to decide.  8 years ago maybe.  But now?  It's been left far too late.

AROS has been badly managed, and heading in various wonky-arsed directions since the start.  The position it is in today is possibly the worst it could be.  Unable to run any of the software that the platform is loved for, while at the same time hamstrung by the same 1980's limitations.  Comparisons to a chocolate teapot would not be completely unfair.

I mean no disrespect to Rob, but I don't think he quite gets the whole Amiga thing, and although I'd welcome a modern Amiga-like OS, I don't think he will be able to deliver.  He admits himself that he is at heart a *nix hacker, and I think this is reflected in his ideas about a new breed AROS, which reads very much like yet another *nix  system.  

To be fair there seems to be a similar movement from within the Amiga community itself, usually presenting itself as 'porting' syndrome, often in the form of "why don't you port X, so we can have ability Y".  All that will lead to is a {bleep} Ami-nix mess that will satisfy no-one.

If someone could create a system where at user level (and by that I mean everything down to the File structure) was as close to AmigaOS 3+ as possible even though it was using existing open source code underneath.  By this I mean something where all the commands in C: that I'd expect are there and behave as their 3.x counterparts, where Graphics.library is really Cairo, but I wouldn't know, where CUPS is used, but I only see printer.device, and I can install drivers by dragging .printer files to devs:printers.  As soon as you start looking at etc folders, cryptic commands, editing bizarre configuration files, and XWindows, you might as well be using linux.

IMO AROS would be best placed to dump the x86 side altogether and re-position itself on 68k.  That might sound backwards, but I think being able to provide an open OS for emulation and classic hardware is much more useful, and more likely to enjoy success in the long run.

Sorry if the above seems mean-spirited.  I honestly respect the amount of work that has been put into AROS, but I must talk honestly about how bad the situation is.
 

Offline Belial6

  • Hero Member
  • *****
  • Join Date: Mar 2002
  • Posts: 568
    • Show only replies by Belial6
    • http://www.glasshead.net
Re: aros fork?
« Reply #19 on: February 06, 2008, 09:49:00 PM »
I kind of agree with you.  One of the biggest problems with the Amiga community right now is Amiga Inc..  They own Workbench and Kickstart.  We have new Amiga compatible hardware.  That problem has been taken care of.  We just have now way of running our old software.  If AROS were made to run on the 68k and be fully binary compatible with AmigaOS, we would have a solid place to start, and the term 'Amiga like' could be replaced with 'Aros like'.  This would allow the community to cut the line to the anchor that is Amiga Inc.

From there, upgrades and rewrites could happen in a more useful fashion.  The biggest thing about any future incompatible version of AROS is the ability for the OS to identify the version of an application that is required to run the code.

As far as I know, to date every platform leaves it to the application to make sure that it is compatible with the OS it runs on.  This means that when the application is abandoned, (as most Amiga software is) it quickly becomes unusable on new versions of the OS.

If the OS asks the application what version it runs on, it can sandbox anything that is no longer compatible, and seamlessly run it in a compatible environment.
 

Offline bloodline

  • Master Sock Abuser
  • Hero Member
  • *****
  • Join Date: Mar 2002
  • Posts: 12113
    • Show only replies by bloodline
    • http://www.troubled-mind.com
Re: aros fork?
« Reply #20 on: February 06, 2008, 10:28:13 PM »
Quote

Belial6 wrote:
I kind of agree with you.  One of the biggest problems with the Amiga community right now is Amiga Inc..  They own Workbench and Kickstart.  We have new Amiga compatible hardware.  That problem has been taken care of.  We just have now way of running our old software.  If AROS were made to run on the 68k and be fully binary compatible with AmigaOS, we would have a solid place to start, and the term 'Amiga like' could be replaced with 'Aros like'.  This would allow the community to cut the line to the anchor that is Amiga Inc.

From there, upgrades and rewrites could happen in a more useful fashion.  The biggest thing about any future incompatible version of AROS is the ability for the OS to identify the version of an application that is required to run the code.

As far as I know, to date every platform leaves it to the application to make sure that it is compatible with the OS it runs on.  This means that when the application is abandoned, (as most Amiga software is) it quickly becomes unusable on new versions of the OS.

If the OS asks the application what version it runs on, it can sandbox anything that is no longer compatible, and seamlessly run it in a compatible environment.


The AROS source code is there... if anyone wanted to they could take it and get it compling on 68K... Bernd is the only one who has so far with his AFA... that sorce code is not going anywhere, It just takes someone with the motivation to do it...

Offline itix

  • Hero Member
  • *****
  • Join Date: Oct 2002
  • Posts: 2380
    • Show only replies by itix
Re: aros fork?
« Reply #21 on: February 06, 2008, 10:49:15 PM »
Quote

Can you open WinUAE/AmigaOS windows as host (Windows) windows ? screens ? not even these ? how does it compare then ?


I dont know. But AROS2 with EUAE integration can not, because none of those two yet exist.

Quote

Depends what your definition of "integration" is, for me it's files, windows, screens. So, why could this not be on top of an API incompatible OS ? files are already integrated in WinUAE at least, sure it would be more code invloved than on a API compliant or -like system, but nevertheless, what would be the obstacle ?


So, what is actually this EUAE integration? If one can integrate EUAE into AROS2 which is neither API or source compatible, why one can not integrate EUAE into Linux in the same fashion?

What is so special in EUAE integration that it could not be done in other systems? And how one can guarantee it works in the system which does not yet exist.

Quote

If you had a blind cousin and if he got his vision back but suddennly and unfortunately became deaf would you not still regard him your cousin ? you see, you would (learn to) communicate differently with him now, in an "incompatible" way.


It is more like that blind cousin would be both blind and deaf.

There is nothing wrong in AROS2 idea and it is very good one. I just hope Amiga (community) does not ruin it. AROS is  already sidetracked to PPC and UAE integration ideas.

Welcome to the ship, AROS. Together we stand, and together we sink.
My Amigas: A500, Mac Mini and PowerBook
 

Offline Einstein

  • Sr. Member
  • ****
  • Join Date: Dec 2004
  • Posts: 402
    • Show only replies by Einstein
Re: aros fork?
« Reply #22 on: February 06, 2008, 11:33:10 PM »
Quote

itix wrote:
Quote

Can you open WinUAE/AmigaOS windows as host (Windows) windows ? screens ? not even these ? how does it compare then ?


I dont know. But AROS2 with EUAE integration can not, because none of those two yet exist.


That's not an answer to my question.

Quote
So, what is actually this EUAE integration? If one can integrate EUAE into AROS2 which is neither API or source compatible, why one can not integrate EUAE into Linux in the same fashion?

What is so special in EUAE integration that it could not be done in other systems? And how one can guarantee it works in the system which does not yet exist.


I already wrote it *could*, but you should have read my comments above perhaps ?

Guarantee ? I guess we'll have to see.

Quote
Quote

If you had a blind cousin and if he got his vision back but suddennly and unfortunately became deaf would you not still regard him your cousin ? you see, you would (learn to) communicate differently with him now, in an "incompatible" way.


It is more like that blind cousin would be both blind and deaf.


Should I interpret that "AROS 2" would not have an API, no communication with anything, at all ?  :-?

Quote
There is nothing wrong in AROS2 idea and it is very good one. I just hope Amiga (community) does not ruin it. AROS is  already sidetracked to PPC and UAE integration ideas.


PPC I have nothing to say about. E-UAE integration however, I don't understand, explain ?

*EDIT*
Ok you refer to 68k binary compatible OS replacement.
*/EDIT*

Quote
Welcome to the ship, AROS. Together we stand, and together we sink.


*Others* may comment on that :)
I have spoken !
 

Offline Argo

  • Hero Member
  • *****
  • Join Date: Feb 2002
  • Posts: 3219
    • Show only replies by Argo
Re: aros fork?
« Reply #23 on: February 07, 2008, 12:05:34 AM »
I wonder if he has taken a look at DragonFly BSD?
 

Offline HenryCase

  • Hero Member
  • *****
  • Join Date: Oct 2007
  • Posts: 800
    • Show only replies by HenryCase
Re: aros fork?
« Reply #24 on: February 07, 2008, 12:09:45 AM »
@All
I have quite a different vision for AROS than most here, as I don't see the split between AROS1 and AROS2 being necessary. My vision of AROS is AROS Ultimate, i.e. all the benefits of both AROS1 and AROS2.

What I'd like you to tell me is what are the 'sacred cows' of AROS1 and AROS2, i.e. the features that cannot be sacrificed. I'll start...

AROS1:
AmigaOS 3.1 API source compatibility

AROS2:
Full memory protection

Both of those can be achieved in AROS Ultimate. I'll explain how if you're interested. Find me more 'sacred cows'! :-D
"OS5 is so fast that only Chuck Norris can use it." AeroMan
 

Offline Einstein

  • Sr. Member
  • ****
  • Join Date: Dec 2004
  • Posts: 402
    • Show only replies by Einstein
Re: aros fork?
« Reply #25 on: February 07, 2008, 12:26:16 AM »
Quote

HenryCase wrote:

I'll explain how if you're interested. Find me more 'sacred cows'! :-D


I don't know about sacred cows, but there's a "sacred" bacon hiding somewhere around here, you've encountered it :-D
I have spoken !
 

Offline persia

  • Hero Member
  • *****
  • Join Date: Sep 2006
  • Posts: 3753
    • Show only replies by persia
Re: aros fork?
« Reply #26 on: February 07, 2008, 01:49:46 AM »
3.1 compatibility should be at the UAE level.  Too much needs to be tossed out to bring the OS to 2008 standards.  
[SIGPIC][/SIGPIC]

What we\'re witnessing is the sad, lonely crowing of that last, doomed cock.
 

Offline Crumb

  • Hero Member
  • *****
  • Join Date: Mar 2002
  • Posts: 1786
  • Country: 00
    • Show only replies by Crumb
    • http://cuaz.sourceforge.net
Re: aros fork?
« Reply #27 on: February 07, 2008, 12:51:50 PM »
@Einstein

Quote
AROS does *not have* binary compaibility


AROS m68k should have it.

Quote
but when it does (E-UAE)


...or when it does (AROS m68k) :-)

Quote
Please don't start with your usual "integrated (E-)UAE will suck because it's nothing like the emulation layers of MOS/OS4 etc etc", pesonally I don't need anything that integrated, besides I will not use any potential new OS on anything but x86 in the next decade, and if so it might not be big-endian anyway.


Imagine that I drag an Icon from the host OS desktop to an opus5 window... will it copy the file? I doubt it :-)

BTW, have you actually tried OS4/MOS at all? Have you tried "GLUAE"? It's more or less the same kind of launcher the bounty wants to achieve but without the layers patch to show windows with other background.

BTW, a patch exist for EUAE to show the windows on top of Linux ones. Unfortunately it's not very advanced and shows all the windows in the same "layer" so host OS windows can't be between emulated WB windows.

Now let's suppose EUAE integration gets finished and you decide to use something called AROS2. First of all... why not use a Linux kernel or any other kernel like NewOS? Then you get posix compatibility. But automatically all AROS/AmigaOS devices/libraries become incompatible. Who cares anyway since Linux&GNU already has all the drivers we could want, these are actively developed and it also has tons of interesting libraries?

You could modify it to get a directory structure similar to AmigaOS, to boot reading an Startup-sequence file, to store commands on /c instead of /bin you could add amiga style path support, you could add a WB like desktop that avoids using XWindows (or maybe not, maybe you want to run all the GUI on top of XWindows).

Since we agree amiga apps are old and modern linux apps are more useful and interesting there's no sense in keeping graphics.library. We could switch to Cairo for every graphic operation (switching to Cairo would make sense even on current AROS... intuition/graphics could run on top of it). AHI is also outdated, we could use OpenAL instead of it.

See... amiga stuff and API is outdated... there's little you would reuse on a modern OS.


If I started an amiga inspired OS (note I say amiga inspired and not amiga-like) I would choose a kernel like linux or NewOS and try to adapt existing software to run in a similar way as AmigaOS.

Changing the kernel to keep a directory structure similar to AmigaOS wouldn't be difficult.

The problem reusing current AROS stuff is that it wouldn't have an easy way to communicate with the new OS. There's no much difference between standard OS libraries/components/devices and third party ones (the exception may be exec/dos/graphics/intuition/layes). That would cause that current AROS sources would be hard to adapt.

Quote
What is amiga like ?


When I say Amiga-like I mean OSes that work the same way as AmigaOS. Just like when I say Unix-like I expect the OS to include a set of posix functions, to have similar commands and I expect to code all unix-like OSes in more or less the same way.

I'm not refering just to the end-user view.

Quote
Actually in case Rob forks it, he will not include anything "amiga" in it's name, he made that clear. And btw, you contradict your self here, look below (look for "LOOK HERE").


What I'm trying to say is that if you get rid of the amiga/aros API why call your fork amiga or aros? Or why show it as successor of amiga/aros if it's not related to it (just using a similar GUI in the first versions?). GEM and MacOS looked and were used in a similar way but they were not related.

IMHO Rob should fork. But there's little stuff that can be reused. I gave him some suggestion: design a new API and provide a library to be used on new Aros programs written for current AROS. Advice coders to stop using amigaos functions and provide them your functions. It's similar as if we were leaving amigaos and jumping to unix, you would advice coders to start using GeekGadgets. Just like that, he would define functions to do message-passing and other amiga-API stuff and AROS coders could start to migrate their code. Once most of apps and maybe libraries were adapted he would at least have something from AROS to use. Anyway since most of AROS stuff is based on old stuff and old APIs you could perfectly start from scratch changing the intuition/graphics calls by Cairo calls and stuff like that.

In conclusion: Fork AROS? Of course, but since everyone agrees that AmigaOS3.1 API is old and outdated why base your new OS on that?


Quote
There's MOS, OS4, and AROS.. but wait, AROS is not "amiga" according to many amiga zealots anyway.


For me AROS/OS4/OS3/MOS are "amiga" :-) I may like some solutions more than the others but I like them all.

Quote
Users don't care crap about the internals, programmers might, *intelligent* programmers will not.


Intelligent programmers that want to write an OS without the limitations of OS3.1 won't base his code on OS3.1 compatible code.

Users don't care about internals and that's the reason they shouldn't discuss internals of OSes.

Quote
if one breaks backwards compatibilty at soure/binary level then it's "anything but amiga-like"


You are right. Other things may look similar or use a similar GUI, but wouldn't be amiga-like. Just like running Amiga-E and a Zune clone on Linux won't make your linux box amiga-like, even if you have put a nice wb-like and even if you rename your /bin as /c and even if you create aliases so you can type "dir" instead of "ls". That's merely cosmetical.



Quote
if you happen to *know* that *nothing* in the old API *implementation* can be reused (simetimes with modification) than why don't you just point it to him with *detail*, you seem to care alot i mean


Detailed suggestion:
-take linux/bsd/newos kernel and modify it if you need it
-use as much standard stuff as you can and avoid using OS3.1 code (that means avoid using AROS code)
-change the OS to use a similar structure to amigaos
-put EUAE and add a launcher for ADF files (most of people who don't care about OS4/MOS-like emulation integration only remembers playing games on A500 in their childhood and haven't touched an amiga for years so they won't miss any program)
The only spanish amiga news web page/club: Club de Usuarios de Amiga de Zaragoza (CUAZ)
 

Offline bloodline

  • Master Sock Abuser
  • Hero Member
  • *****
  • Join Date: Mar 2002
  • Posts: 12113
    • Show only replies by bloodline
    • http://www.troubled-mind.com
Re: aros fork?
« Reply #28 on: February 07, 2008, 02:29:12 PM »
What I think everyone is missing here is not suggesting throwing away AROS and the associated systems and structures... What I think we are suggesting is where we need to modernize something and it will break compatibility then we should go for it... At the moment the default is to maintain compatibility even if that means ugly hacks to include new features.

Offline downix

  • Hero Member
  • *****
  • Join Date: Jan 2003
  • Posts: 1587
    • Show only replies by downix
    • http://www.applemonthly.com
Re: aros fork?
« Reply #29 from previous page: February 07, 2008, 02:35:18 PM »
Rather than fork why not split?

See, the whole EUAE for legacy bit suffers from one big shortfall:  No legal ROM/Workbench

So, focus one branch of the split to being this ROM/Workbench, focusing all efforts on delivering maximum compatibility.  

Now your new branch can progress forwards, breaking compat where needed, without worries.
Try blazedmongers new Free Universal Computer kit, available with the GUI toolkit Your Own Universe, the popular IT edition, Extremely Reliable System for embedded work, Enhanced Database development and Wide Area Development system for telecommuting.