Amiga.org

Operating System Specific Discussions => AROS Research Operating System => Topic started by: HenryCase on January 21, 2008, 08:14:17 PM

Title: New AROS bounties
Post by: HenryCase on January 21, 2008, 08:14:17 PM
Hi all,

Just wondering which of the AROS bounties you'd like to be tackled next, and also whether anyone had any ideas of brand new AROS bounties they'd like to see.

Looking at the current bounties (http://www.thenostromo.com/teamaros2/), I'd say the most important unassigned bounties are those for Kickstart Replacement, ACPI (Phase I) and SoundBlaster AHI Driver Bounty (in that order IMHO).

As for new bounties, there are so many good possibilities. One inspiration for this thread was reading Rob N's great Traveller blog (http://cataclysm.cx/). In the comments section of the most recent post (18/01), someone mentioned that Gnash, the FOSS Flash player, can use Cairo to display its graphics. A bounty for this would be awesome, as it would enhance the work done on Traveller (I admit I spend quite a lot of time on YouTube which is why I'd want an AROS port, but other websites would benefit too).

I'd also like to see bounties created to support those who do more of the back end, lower level AROS stuff, as they are really helping the stability and usefulness of the OS. What these bounties should be I don't know. Can someone here tell us which low level sections of AROS need most work?

I'd also like to see a bounty being started to support the AROS mp3 player AROSamp, as its at an early stage right now and I'm hoping more money equals faster development.

A bounty for more integrated ARexx support would also be awesome.

That will do from me for now. Please post your AROS bounty ideas.
Title: Re: New AROS bounties
Post by: yorgle on January 21, 2008, 08:44:58 PM
Personally, I'd love to see the AROS running on OS X bounty fulfilled.  I understand that porting AROS to other platforms should be considered after there's more functionality to port, but I personally believe that if it is running on more platforms, it has a better chance of pulling in more developers to help get the other functionality working, or apps ported and so on.

I don't really have a linux box or a useful PC, but I have many Macs, and although one of my apps was ported by someone else at one point (UberPaint), I'd love to be able to do that work myself.  

I have new directions I'm taking UberPaint and other apps I'm designing currently... and yes, they will be ported to AROS eventually... I'd just love to be able to be doing it sooner myself. ;)

As far as other AROS bounties, I'd love to see native PPC support running on Mac hardware.  There are a lot of old powerbooks around that would make great AROS boxes. ;)
Title: Re: New AROS bounties
Post by: HenryCase on January 21, 2008, 10:05:56 PM
@yorgle
Well I'm all up for making sure potential application developers have an easy time running AROS, but have got to be honest in saying that I see the platform ports as a low priority right now.

Can you run AROS hosted on PPC Linux? Or run VmwAROS? Until that OSX bounty is fulfilled I mean, I'd like to see it happen too, I just think other parts of AROS should be given focus first. All IMO of course.

How about brand new bounties? Any coding resources you'd like to see implemented?
Title: Re: New AROS bounties
Post by: Krusher on January 21, 2008, 10:24:50 PM
A simple programming language, like Amos. The sources are out there, so it should be not the biggest problem getting this on Aros.

I loved Amos as a programming language. It was the first and last one I did something useful with, I wrote a remote control program, using the serial port so I could control the Amiga in various ways elsewhere, be it on a Terminal emulation or a real Terminal (I had a Wyse)

Basically, an operating system needs a programming language that's accessible for everyone.

Next, but that's already covered, a decent Internet browser.
Title: Re: New AROS bounties
Post by: bloodline on January 21, 2008, 10:39:30 PM
I think something like  Mattahtias BASIC would probably be a good bet...

http://sourceforge.net/projects/mattathias/
Title: Re: New AROS bounties
Post by: Krusher on January 21, 2008, 10:45:17 PM
My best bet is getting people the programming language they remember.
Title: Re: New AROS bounties
Post by: bloodline on January 21, 2008, 10:53:29 PM
Quote

Krusher wrote:
My best bet is getting people the programming language they remember.


Yes... and Mattahtias BASIC is an open source AMOS clone... that's why I suggested it :-)
Title: Re: New AROS bounties
Post by: Krusher on January 21, 2008, 10:55:30 PM
Oops, did not know that, sorry.

But why use that since the sourcecode of Amos is published?
Title: Re: New AROS bounties
Post by: bloodline on January 21, 2008, 11:01:02 PM
Quote

Krusher wrote:
Oops, did not know that, sorry.

But why use that since the sourcecode of Amos is published?


AMOS is totally 68k ASM and tied completely to the Amiga hardware.
Title: Re: New AROS bounties
Post by: Krusher on January 21, 2008, 11:09:24 PM
Quote

bloodline wrote:
...
AMOS is totally 68k ASM and tied completely to the Amiga hardware.


Good for me as a hardcore lover of Amiga hardware.

New Bounty then, 68k ASM compatible. But that's already a bounty right?
Title: Re: New AROS bounties
Post by: bloodline on January 21, 2008, 11:24:31 PM
Quote

Krusher wrote:
Quote

bloodline wrote:
...
AMOS is totally 68k ASM and tied completely to the Amiga hardware.


Good for me as a hardcore lover of Amiga hardware.

New Bounty then, 68k ASM compatible. But that's already a bounty right?


Yes, the AROS - Kickstart replacement bounty... that's one I'm really looking forward to!
Title: Re: New AROS bounties
Post by: Einstein on January 22, 2008, 12:09:56 AM
Hi HenryCase!

Quote
What these bounties should be I don't know. Can someone here tell us which low level sections of AROS need most work?


Resource managers (for resource tracking), including memory manager (for memory tracking and protection).

Result: criminal tasks, instead of the OS, will experience their afterlife. :-)
Title: Re: New AROS bounties
Post by: HenryCase on January 22, 2008, 12:40:32 AM
@Krusher
Quote
Basically, an operating system needs a programming language that's accessible for everyone.

I couldn't agree more. Programs like AMOS and Blitz Basic seemed reasonably popular back in the day, hopefully Mattahtias BASIC will be too. I've got a couple of ideas for simple apps I'd like to write using it already.

@bloodline
Quote
Yes, the AROS - Kickstart replacement bounty... that's one I'm really looking forward to!

I think I've found a way that the creation of a Kickstart replacement (which is definitely number one in terms of importance out of the unassigned bounties) can be sped up. Have you ever heard of how IBM-PC clones started? Let me explain...

One of the key proprietary parts of the IBM PC back before the clones came was the BIOS. However, Phoenix Technologies (creator of the first IBM PC BIOS clone) worked out an ingenious way of getting around the reverse engineering problem, by using the 'clean room' technique. From Wikipedia:

"With the success of the IBM PC in 1983, Phoenix decided to provide a IBM PC compatible ROM BIOS to the PC market. A licensable ROM BIOS would allow clone PC manufacturers to run the same applications, and even the MS-DOS that was being used by IBM. However, to do this Phoenix needed a strategy for defense against IBM copyright infringement lawsuits. IBM would claim that the Phoenix programmers had copied parts of the IBM BIOS code published by IBM in its Technical Reference manuals. Because, due to the nature of low-level programming, in two well-written pieces of code that perform the same function some correspondence is inevitable, it would be impossible for Phoenix to defend itself on the grounds that no part of its BIOS matched IBM's. Phoenix developed a "clean room" technique that isolated the Engineers who had been contaminated by reading the IBM source listings in the IBM Technical Reference Manuals. The contaminated Engineers wrote specifications for the BIOS APIs and provided the specifications to "clean" Engineers who had not been exposed to IBM BIOS source code. Those "clean" Engineers developed code from scratch to mimic the BIOS APIs. This technique provided Phoenix with a defensibly non-infringing IBM PC-compatible ROM BIOS. Because the programmers who wrote the Phoenix code had never read IBM's, nothing they wrote could have been copied from IBM's code, no matter how closely the two matched. The first Phoenix PC ROM BIOS was introduced in May, 1984, and helped fuel the growth in the PC industry."

Now the Kickstart is low level code and can be reverse engineered. If we can think of a way to recreate the clean room technique and assign two or more 68k coders to assist each other in this way then we're on to a winner (there also needs to be a fair way to split the bounty but lets not worry about that right now).

The only way I can think of to recreate the 'clean room' technique through the Internet is if the person doing the decompiling and analysis stays completely anonymous and unreachable to the 68k code writers. Or maybe there is another way to do it. Any ideas?

Quote
Einstein wrote:
Hi HenryCase!

Quote
What these bounties should be I don't know. Can someone here tell us which low level sections of AROS need most work?


Resource managers (for resource tracking), including memory manager (for memory tracking and protection).

Result: criminal tasks, instead of the OS, will experience their afterlife. :-)


Hi Einstein! :-)

Yep, they sound suitably important for a bounty. Apart from memory what other resources would need managers? Also, doesn't the x86-64 version of AROS already have partial memory protection? If so, is there any reason why this can't be applied to the other versions of AROS fairly quickly?
Title: Re: New AROS bounties
Post by: bloodline on January 22, 2008, 01:01:32 AM
Quote

HenryCase wrote:

Now the Kickstart is low level code and can be reverse engineered. If we can think of a way to recreate the clean room technique and assign two or more 68k coders to assist each other in this way then we're on to a winner (there also needs to be a fair way to split the bounty but lets not worry about that right now).

The only way I can think of to recreate the 'clean room' technique through the Internet is if the person doing the decompiling and analysis stays completely anonymous and unreachable to the 68k code writers. Or maybe there is another way to do it. Any ideas?


I can assure you the legal issues of this and all techniques required have been discussed in great detail on the AROS dev list. The only problem is lack of interested developers.
Title: Re: New AROS bounties
Post by: HenryCase on January 22, 2008, 01:17:53 AM
Quote

bloodline wrote:
Quote

HenryCase wrote:

Now the Kickstart is low level code and can be reverse engineered. If we can think of a way to recreate the clean room technique and assign two or more 68k coders to assist each other in this way then we're on to a winner (there also needs to be a fair way to split the bounty but lets not worry about that right now).

The only way I can think of to recreate the 'clean room' technique through the Internet is if the person doing the decompiling and analysis stays completely anonymous and unreachable to the 68k code writers. Or maybe there is another way to do it. Any ideas?


I can assure you the legal issues of this and all techniques required have been discussed in great detail on the AROS dev list. The only problem is lack of interested developers.


Well I'm glad the legal issues surrounding the techniques have been discussed. As for the lack of willing/interested developers, as far as I can tell it is a huge task, and one where a team would be preferable. I am prepared to dig around some places we may find willing coders if I knew these things:

1. How are bounties split when there is a team (or if this hasn't happened in the past has this been discussed)?
2. Other than the clean room technique, what other legal methods of creating a kickstart replacement are there?
3. Was there a strong consensus on which would be the best (or easiest) method to use?

Other than the kickstart replacement bloodline, any AROS bounties you'd like to see?
Title: Re: New AROS bounties
Post by: zizban on January 22, 2008, 01:23:18 AM
How about screen refresh rates in the current ati/nvidia drivers? Not all of us use LCDs and a low refresh rate on a CRT sucks.
Title: Re: New AROS bounties
Post by: alenppc on January 22, 2008, 01:26:39 AM
Quote

Other than the kickstart replacement bloodline, any AROS bounties you'd like to see?


Port AROS to the PS3... time to get rid of the damn penguin. :-)

But this one will always be just a dream though.
Title: Re: New AROS bounties
Post by: Einstein on January 22, 2008, 01:37:56 AM
Quote

HenryCase wrote:

Hi Einstein! :-)

Yep, they sound suitably important for a bounty. Apart from memory what other resources would need managers?


Anything that is a resource has it's manager, e.g: memory (Exec), message ports (Exec), windows (Intuition), Zune/MUI objects (Zune/MUI, this includes BOOPSI too), AROS' Traveller/Cairo and beyond (Robert Norris / fce2) :lol: and alot more..

Quote
Also, doesn't the x86-64 version of AROS already have partial memory protection? If so, is there any reason why this can't be applied to the other versions of AROS fairly quickly?


Difference between partial mp (A) and full mp (B) is like being attacked by a samurai when one carries a shield (A) and when one sits in a tank (B). :-D
Partial mp is just a retrofit mp, with full mp huge portion of the API needs modification/rewrite, that's why it takes time.
As for AROS32, i don't know if that will be supported for enhancements, it's up to the devs.
Title: Re: New AROS bounties
Post by: HenryCase on January 22, 2008, 01:38:38 AM
@zizban
Yes, that sounds essential. I wonder how hard that would be to implement.

@alenppc
PS3 AROS. Pipe dream, or is it...
http://www.geek.com/mit-offers-free-playstation-3-programming-course/
http://www.cag.csail.mit.edu/ps3/lectures.shtml
Get yer learnin' hat on! :-D

@Einstein
Quote
Anything that is a resource has it's manager, e.g: memory (Exec), message ports (Exec), windows (Intuition), Zune/MUI objects (Zune/MUI, this includes BOOPSI too), and alot more..

Do you mean to say these resource managers don't exist yet or that they need a makeover?

Quote
Difference between partial mp (A) and full mp (B) is like being attacked by a samurai when one carries a shield (A) and when one sits in a tank (B). :-D

Liking the analogy there Einstein! :-D

Quote
Partial mp is just a retrofit mp, with full mp huge portion of the API needs modification/rewrite, that's why it takes time.
As for AROS32, i don't know if that will be supported for enhancements, it's up to the devs.

I see. Well in that case I'd prefer if we had full MP (sounds like were playing a FF game!) on one distro before work was done to port it to all of them, but as you said it's up to the devs.
Title: Re: New AROS bounties
Post by: bloodline on January 22, 2008, 01:47:16 AM
Quote

Einstein wrote:

Quote
Also, doesn't the x86-64 version of AROS already have partial memory protection? If so, is there any reason why this can't be applied to the other versions of AROS fairly quickly?


Difference between partial mp (A) and full mp (B) is like being attacked by a samurai when one carries a shield (A) and when one sits in a tank (B). :-D
Partial mp is just a retrofit mp, with full mp huge portion of the API needs modification/rewrite, that's why it takes time.
As for AROS32, i don't know if that will be supported for enhancements, it's up to the devs.


I'm pretty sure that any enhancements that can be brought in from AROS64 into the other ports will be. x86 is currently being finalised for ABI V1 by the workaholic Staf, I expect him, Robert and Michal will sort something out before ABI V1 is finished... and there are bunch of guys working on the PPC port, which is nowhere near a finalised ABI (but also has Michal Schulz working on it) so plenty of time for it to be improved.

:-)
Title: Re: New AROS bounties
Post by: SamuraiCrow on January 22, 2008, 01:52:09 AM
Quote

bloodline wrote:
Quote

Krusher wrote:
My best bet is getting people the programming language they remember.


Yes... and Mattahtias BASIC is an open source AMOS clone... that's why I suggested it :-)


Mattathias BASIC will be more than a clone.  It will be more like a backward compatible sequel.  Imagine Amos 3.0 with AGA support or graphics card support running in any machine language you would like.  8-)
Title: Re: New AROS bounties
Post by: bloodline on January 22, 2008, 01:52:21 AM
Quote

HenryCase wrote:

Well I'm glad the legal issues surrounding the techniques have been discussed. As for the lack of willing/interested developers, as far as I can tell it is a huge task, and one where a team would be preferable. I am prepared to dig around some places we may find willing coders if I knew these things:

1. How are bounties split when there is a team (or if this hasn't happened in the past has this been discussed)?


Never been done before, the team members will have to figure it before they start work, and then inform Dammy.

Quote

2. Other than the clean room technique, what other legal methods of creating a kickstart replacement are there?


Firstly lets get something that POSTs the machine :-) all that requires is a Hardware Reference Manual and plenty of time/motivation... we make it compatible later ;-)

Quote

3. Was there a strong consensus on which would be the best (or easiest) method to use?


Yes... we need more 68k, Amiga hardware hacker devs. :-)

Quote

Other than the kickstart replacement bloodline, any AROS bounties you'd like to see?


Yes, actually... I want EFI and SMP support... but I can only dream at the moment :-)

-Edit- Oh and S-ATA support would be nice too... but there is a bounty for that already.
Title: Re: New AROS bounties
Post by: Einstein on January 22, 2008, 01:52:35 AM
Quote

bloodline wrote:
Quote

Einstein wrote:

Quote
Also, doesn't the x86-64 version of AROS already have partial memory protection? If so, is there any reason why this can't be applied to the other versions of AROS fairly quickly?


Difference between partial mp (A) and full mp (B) is like being attacked by a samurai when one carries a shield (A) and when one sits in a tank (B). :-D
Partial mp is just a retrofit mp, with full mp huge portion of the API needs modification/rewrite, that's why it takes time.
As for AROS32, i don't know if that will be supported for enhancements, it's up to the devs.


I'm pretty sure that any enhancements that can be brought in from AROS64 into the other ports will be. x86 is currently being finalised for ABI V1 by the workaholic Staf, I expect him, Robert and Michal will sort something out before ABI V1 is finished... and there are bunch of guys working on the PPC port, which is nowhere near a finalised ABI (but also has Michal Shulz working on it) so plenty of time for it to be improved.

:-)


Well something tells me it would take time anyway  :-P
Title: Re: New AROS bounties
Post by: bloodline on January 22, 2008, 01:58:37 AM
Quote

SamuraiCrow wrote:
Quote

bloodline wrote:
Quote

Krusher wrote:
My best bet is getting people the programming language they remember.


Yes... and Mattahtias BASIC is an open source AMOS clone... that's why I suggested it :-)


Mattathias BASIC will be more than a clone.  It will be more like a backward compatible sequel.  Imagine Amos 3.0 with AGA support or graphics card support running in any machine language you would like.  8-)


I hoped you'd show up :-) Come on lets get Mattahtias Basic on AROS! :-D
Title: Re: New AROS bounties
Post by: Einstein on January 22, 2008, 02:03:28 AM
@HenryCase

Quote

Do you mean to say these resource managers don't exist yet or that they need a makeover?


A makeover.

Quote
Partial mp is just a retrofit mp, with full mp huge portion of the API needs modification/rewrite, that's why it takes time.
As for AROS32, i don't know if that will be supported for enhancements, it's up to the devs.

I see. Well in that case I'd prefer if we had full MP (sounds like were playing a FF game!) on one distro before work was done to port it to all of them, but as you said it's up to the devs.[/quote]

Or up to the bounty cash  :-)
Title: Re: New AROS bounties
Post by: HenryCase on January 22, 2008, 02:14:15 AM
@SamuraiCrow
Mattathias BASIC does sound great. I posted a few questions on the project's SourceForge page, hope you don't mind me repeating them here:
1. How far through the development is this project (i.e. is a release candidate or beta version due soon)?
2. Which version of AmigaOS is going to be your first target platform (please say AROS!)?
3. Does Mattathias BASIC follow similar syntax to AMOS?

@bloodline
Thanks for the answers to my kickstart coders questions, I'll get chasing them tomorrow. Didn't know about EFI (BIOS replacement). Wouldn't it be better to get UEFI support considering thats the newer standard, or did you want it for x86 Mac compatibility? SMP support is a must, that should definitely be a bounty (maybe one to start next year, but collecting funds should start ASAP).

As for SATA, isn't the first SATA driver currently in progress (for the Sam440 bounty)? If so, will this reduce the workload in writing other SATA drivers?

Quote
Einstein wrote:
A makeover.


I see. Well, if it makes AROS more stable then I'm all for it (not that I'm saying it isn't stable now you understand!).

Quote
Einstein wrote:
Or up to the bounty cash  :-)


Yeah, I can see that one being real popular! 'Pay AROS devs to give you partial MP only for them to perform a huge rewrite when full MP is needed'. :-D
Title: Re: New AROS bounties
Post by: XDelusion on January 22, 2008, 02:56:17 AM
I'd be interested in seeing it on the XBOX. 64Mb of RAM would go a long way with an Amiga like OS I'd think.
Title: Re: New AROS bounties
Post by: bloodline on January 22, 2008, 09:30:36 AM
Quote

XDelusion wrote:
I'd be interested in seeing it on the XBOX. 64Mb of RAM would go a long way with an Amiga like OS I'd think.


For the consoles, I think the best option is to run AROS on top of a minimal Linux, that would solve pretty much all problems and be quick to implement.
Title: Re: New AROS bounties
Post by: Piru on January 22, 2008, 11:38:57 AM
@HenryCase
Quote
reverse engineering & reimplementing Kickstart ROM

This is a huge task, however. You make it sound as if it'd be couple of months for 2-3 guys.

More like 5-10 years, assuming you find skilled and well motivated hackers to do it.

PC BIOS was only couple of KB, trivial in comparison. Yet it took 15 guys several months or so.
Title: Re: New AROS bounties
Post by: jj on January 22, 2008, 12:16:40 PM
I would imagine that the Kickstart roms will become public domain , free to use before a replacement is coded
Title: Re: New AROS bounties
Post by: bloodline on January 22, 2008, 01:20:51 PM
Quote

Piru wrote:
@HenryCase
Quote
reverse engineering & reimplementing Kickstart ROM

This is a huge task, however. You make it sound as if it'd be couple of months for 2-3 guys.

More like 5-10 years, assuming you find skilled and well motivated hackers to do it.

PC BIOS was only couple of KB, trivial in comparison.


Indeed, for AROS the real quest is to write a ROM that will POST the machine and allow it to boot AROS... That should be enough for most software (OS legal apps would be dependant on AROS, and most games and demos tend to bypass the ROM once the machine is brought up to a known state)... then work can progress at making it more compatible if needs must.
Title: Re: New AROS bounties
Post by: Einstein on January 22, 2008, 11:48:03 PM
Quote

HenryCase wrote:

I see. Well, if it makes AROS more stable then I'm all for it (not that I'm saying it isn't stable now you understand!).


I understand, but the samurai..  :evilgrin:
Title: Re: New AROS bounties
Post by: Mangar on January 23, 2008, 12:24:06 AM
I think SATA support is really important since most modern systems use SATA HDD's.
Title: Re: New AROS bounties
Post by: HenryCase on January 23, 2008, 01:10:59 AM
Quote
Piru wrote:
@HenryCase
Quote
reverse engineering & reimplementing Kickstart ROM

This is a huge task, however. You make it sound as if it'd be couple of months for 2-3 guys.

More like 5-10 years, assuming you find skilled and well motivated hackers to do it.

PC BIOS was only couple of KB, trivial in comparison. Yet it took 15 guys several months or so.


Piru, I now understand the enormity of this task...

I downloaded Amiga Kickstart ROMs tonight and tried decompiling one into 68k ASM. The resultant text file is 3406 pages long in Word (using 10pt Courier New). However, it isn't a task we can shy away from, as it isn't just important for AROS but for a number of other projects as well.

Looks like I'm going to be busy building an 68k ASM coder army! :-o :-D

Piru, what new bounties would you like to see for AROS?
Title: Re: New AROS bounties
Post by: bloodline on January 23, 2008, 01:39:12 AM
@HenryCase

Unless you got the offset right you have now got 3000 pages of garbage :-)

By the way you have now rendered youself unable to work on a kickstart replacement :-(
Title: Re: New AROS bounties
Post by: HenryCase on January 23, 2008, 02:14:54 AM
@bloodline
I used a great 68k disassembler called IRAPC, created easily readable ASM code. I'm no coder, but I do recognise the format the code takes. Whether it compiles again into a useable ROM is another matter!

I am really not worried that I cannot work on the kickstart replacement directly, but I would be willing to coordinate the work, as well as try to analyse the structure of the kickstart to assist the coders.

Question is how long will this work take? Lets say I managed to find 10 willing 68k ASM coders, we'd still be looking at 2-3 years before completion. I'd hope that the bounty would have increased quite a bit by this time.
Title: Re: New AROS bounties
Post by: SamuraiCrow on January 23, 2008, 03:31:26 AM
Quote

HenryCase wrote:
@SamuraiCrow
Mattathias BASIC does sound great. I posted a few questions on the project's SourceForge page, hope you don't mind me repeating them here:
1. How far through the development is this project (i.e. is a release candidate or beta version due soon)?
2. Which version of AmigaOS is going to be your first target platform (please say AROS!)?
3. Does Mattathias BASIC follow similar syntax to AMOS?

1.  When the first beta is released, keep in mind that it won't have all of the Amos features added.  We're basing it on LLVM so it will be ready several months if not years sooner than if we wrote the whole thing ourselves.  And to answer your question, it isn't ready yet and probably not too soon.

2.  The AROS version will directly follow the Linux, Windows, and Mac versions.  So yes, it will be the first Amiga-like OS to have Mattathias.

3.  Yes.  That's why it's taking so long.  We are improving the extension format so that a Mattathias BASIC programmer can develop their own extensions in Mattathias.