AROS68K development methodology and bounties

[wXR]

Hello everyone,

I wanted to take this opportunity to separate out of the Critical components to open source thread the lively AROS-related discussion that kicked up there, in order that the thread may return to its primary topic.

As one forum member rightly pointed out, while we work to address the injustice that is the closed-source Amiga OS 3.1, there is a very worthy project called AROS that certainly deserves our attention. To my mind, this is especially true considering the Vampire team’s recent moves, which I see as a boon to awareness and development of the project. It is something that I at least would like to seize upon and assist along in any way possible.

Before the topic of bounties is directly addressed however, I want to express a personal opinion about why they have largely failed up to this point. Some would say that it is a matter of lacking interest, and maybe that is so. But on the whole I take a different view: I believe that the user experience around bounties and AROS development is appallingly bad. Just take a look. As it stands, discussion, development, and bounties are spread out across multiple domains, most of which are running early 2000s-era software. This breaks the continuity completely. For example, here is one bounty that is reasonably funded, and arguably pretty important, yet unclaimed for over five years. Perhaps this is a simple lack of interest, but I suspect UX is as much to blame, if not more so. After all, how many times a day are people checking that domain?

By contrast, here is a bounty that is directly connected with an issue that is able to be engaged very much like a forum thread. Being hosted on GitHub also means that it is accessible (though by no means automatically visible) to all of the existing users of that platform. In other words, GitHub has the potential to act as connective tissue between discussion, development, and bounties. There are also bounty platforms which integrate directly with GitHub to make this aspect even more flexible. On the whole, assuming the goal of creating an active, open community around a project – one which feels inclusive and organized (albeit in the sometimes chaotic style of an active open source project) – I believe that this is the correct way to do things.

Now, getting to the meat of it: I am willing to donate an initial $25,000 to AROS68K-related (ABIv1) bounties, if a team is willing to organize in the manner described above, on GitHub. Yes, my offer requires some initially uncomfortable changes, and perhaps a grumbling abandonment of certain intellectual investments, but I make this offer with a sincere belief that it will be better for everyone in the long run. There are no other strings attached. In fact, I turn everything past this point over to the community: I would very much appreciate it if what came out of this thread, was a lively discussion outlining what you guys see as the most important developments for AROS as a core, standalone operating system for 68K systems – past, present, and future. I will contribute to it myself, but where the bounty money goes will have to be directed by collective good sense – and I sincerely hope you have it. This is not the place to request an MP3 player or a browser, for example. At least not yet.

Please try to be productive, helpful, and on-topic. Anti-AROS sentiment is not welcome in this thread at all; please start another one if you want to rant in that direction.

Power to the people? Let's see how it goes.

[clebin]

Awesome stuff. That's incredibly generous. You could try getting in touch with Jason McMullan. It would be a huge boon if he could be attracted back to the project. He achieved a lot with AROS 68k in a short space of time, not to mention writing a P96 driver for Vampire in a matter of a weekend more recently. Even if he isn't available to do the work himself, I'm sure he has a stack of ideas for improvements and optimisations to make AROS a suitable replacement for OS3.x. Thanks again wXR.

[ascp]

Originally Posted by wXR

...
Now, getting to the meat of it: I am willing to donate an initial $25,000 to AROS68K-related (ABIv0) bounties, if a team is willing to organize in the manner described above, on GitHub. ....

.

Amazing!!

[wXR]

@clebin

If you know Jason, would you mind pointing him to this thread? I would very much love to see him taking part in this. If it would be helpful for me to reach out somehow, I can do that, but it probably would be better handled by someone who knows him; I have no existing relationship with him at all.

[wXR]

By the way, it occurred to me after I posted the above, that some resistance to the abandonment of said "intellectual investments" may arise from the fact that some AROS developers are actually developing natively, rather than cross-platform. This might make GitHub-based development more challenging, at least without a little bit of additional tooling. At which point the concern is that said tooling may perform quite poorly on 68K-based systems. Are any AROS developers presently able to speak to that reality, so we can also see how we might adapt to it, assuming it is any kind of problem?

Speaking to solutions a bit early, I know that strim setup a nice SVN -> git -> GitHub toolchain for Hedeon for the Sonnet project. As I understand, this enables Hedeon to develop natively, while still taking advantage of all of the nice features that GitHub provides.

[Locutus]

Theres work on a native AROS Git client ongoing, i haven't checked out yet at what state it is now.

[ross]

Originally Posted by wXR

Now, getting to the meat of it: I am willing to donate an initial $25,000 to AROS68K-related (ABIv0) bounties

Hi wXR, maybe you mean ABIv1...
ABIv0 is not binary compatible with AOS.

ross

[Overflow]

Given Apollo Teams recent focus on AROS as a real alternative, I like to think bounties will have a bigger chance of being fullfilled going forward, espesially if there are developers showing intrest in taking on the tasks.

[wawa]

Originally Posted by ross

Hi wXR, maybe you mean ABIv1...
ABIv0 is not binary compatible with AOS.

thats right. to explain a bit in depth:

abiv0 predates aros kickstart bounty taken finally by skilled developers toni willen and jason mcmullan and amiga-m68k target being brought back from unmaintained. in parallel with 68k port a target has been engage to modernize aros and finally make it conform with initial goals (as i aundewrstand that). this is abiv1.

abiv0 is bing kept around and updated to v1state, where possible and appropriate, for x86 platform backwards compatibility. but it is condidered obsolete. practically all development work is done on abiv1. thats where smp is being implemented.

there is already lot of talk and tension for as soon as possble, and as effective as possible transition to a new default pc platform which appears to be x64 (64bit), possibly with the multicore support.

[wawa]

@wxr

i still dont understand why do you prioritize git. aros has moved to svn not long ago. i think it has been on git before, at least git mirror has been present then and is still available. i think there were reasons for this decission and i dont think its a good idea to constantly move the source around.

svn if pretty fine to use and the web frontend has seen some work and whoever likes to check out the sources with git, can do so from the mirror. thats how jason worked, with his own repo, pushing his changes up to the main svn one.

but im not an experienced coder, so those who know more should speak up.

[wawa]

what concerns jason, i worked with him and toni, testing aros on amiga hardware. everybody "knows" him in aros team. toni perhaps the best. but i can ping him if you like.

i could also ping deadwood, even though he declared himself inactive for the time being just lately.

edit: ok, i notified them both. now its on them, if they want to speak up.

[amiwolf]

I admire anyone willing to put their money where their mouth is, even more so when they don't fold after the first bad hand. As I'm convinced none of the Amiga rights players dare risk the legal house of cards they've created, AROS is probably our best bet.

My contribution to the kickstart bounty wawa mentioned wasn't much but hey, every bit helps yeah? Got the impression Toni didn't enjoy the experience that much though I guess deadlines are never fun. Motivation for any Amiga developer is more than just money, I think.

Good luck with the bounties, whichever form they take.

[Toni Wilen]

It was interesting and fun (and still is) as long as it was/is not too high level: mainly hardware drivers or dos related (dos is always interesting and weird!), gfx.library, especially intuition and other higher level libs/classes/whatever are very uninteresting and also usually they don't just fit nicely with m68k hardware (like planar bitmaps). I am 100% not interested in optimizing that!

Debugging some big program and finally finding out that problem is caused by (for example) some minor header bug (like wrong register parameter or missing declaration that can only affect m68k), is mostly annoying and really boring.

[wawa]

@toni
therefore in an ideal world contibutors of complementary interests would find themselves together. i think you are delivering awesome work.

@wxr
unfortunatelly jason cannot contribute due to professional obligations. i suspected something like that.

[idrougge]

I'd like to interpret wXR's proposal as a bounty for an AmigaOS native Git client.

[wawa]

today i first booted aros on my a4000 after quite some time.
amazingly the speed is usable. in this thread another member of a1k and me posted fotos of a4000/aros68k running usb masstorage, and browing the web with aros owb. device drivers are the genuine 68k drivers delivered with the hardware.
http://www.a1k.org/forum/showthread....65#post1072165

heres my photos, since i suppose you might need to log in on a1k to see them otherwise:
https://ibb.co/cXQAo5
http://ibb.co/ijTrvk
http://ibb.co/dFART5

[Kalamatee]

Originally Posted by wXR

Hello everyone,

I wanted to take this opportunity to separate out of the Critical components to open source thread the lively AROS-related discussion that kicked up there, in order that the thread may return to its primary topic.

As one forum member rightly pointed out, while we work to address the injustice that is the closed-source Amiga OS 3.1, there is a very worthy project called AROS that certainly deserves our attention. To my mind, this is especially true considering the Vampire team’s recent moves, which I see as a boon to awareness and development of the project. It is something that I at least would like to seize upon and assist along in any way possible.

Before the topic of bounties is directly addressed however, I want to express a personal opinion about why they have largely failed up to this point. Some would say that it is a matter of lacking interest, and maybe that is so. But on the whole I take a different view: I believe that the user experience around bounties and AROS development is appallingly bad. Just take a look. As it stands, discussion, development, and bounties are spread out across multiple domains, most of which are running early 2000s-era software. This breaks the continuity completely. For example, here is one bounty that is reasonably funded, and arguably pretty important, yet unclaimed for over five years. Perhaps this is a simple lack of interest, but I suspect UX is as much to blame, if not more so. After all, how many times a day are people checking that domain?

While I understand where you are coming from, there are a number of considerations regarding this that have been discussed on the AROS mailing list.

The bounties are not handled/managed by the AROS dev team - they are handled by a 3rd party that is not directly affiliated to AROS (as nothing that isn't under the aros.org umbrella, is). There are a number of reasons for this including making sure there is no conflict of interest between the donors, or developers, or any issues of misappropriating funds etc. Having an external entity handle this also prevents the "core" devs taking all the bounties, and makes it more necessary to prove the bounty is fulfilled before it is considered completed and payment made.

AFAIK - the main reason is the lack of (capable) devs to take on those bounties, or lack of interest/motivation from the devs who are able. A number of them are to all extents and purposes complete, or almost complete - but the people who have done the work have not wanted to take the bounty, or not felt able to finish in the allotted time.

By contrast, here is a bounty that is directly connected with an issue that is able to be engaged very much like a forum thread. Being hosted on GitHub also means that it is accessible (though by no means automatically visible) to all of the existing users of that platform. In other words, GitHub has the potential to act as connective tissue between discussion, development, and bounties. There are also bounty platforms which integrate directly with GitHub to make this aspect even more flexible. On the whole, assuming the goal of creating an active, open community around a project – one which feels inclusive and organized (albeit in the sometimes chaotic style of an active open source project) – I believe that this is the correct way to do things.

Now, getting to the meat of it: I am willing to donate an initial $25,000 to AROS68K-related (ABIv0) bounties, if a team is willing to organize in the manner described above, on GitHub. Yes, my offer requires some initially uncomfortable changes, and perhaps a grumbling abandonment of certain intellectual investments, but I make this offer with a sincere belief that it will be better for everyone in the long run. There are no other strings attached. In fact, I turn everything past this point over to the community: I would very much appreciate it if what came out of this thread, was a lively discussion outlining what you guys see as the most important developments for AROS as a core, standalone operating system for 68K systems – past, present, and future. I will contribute to it myself, but where the bounty money goes will have to be directed by collective good sense – and I sincerely hope you have it. This is not the place to request an MP3 player or a browser, for example. At least not yet.

As already mentioned that's unlikely to happen (I'm not saying it wont or is impossible) - but there are concerns from the devs point of view of the implications of handling the bounties, etc, themselves - or the appearance they do if it falls under the remit of the AROS.org domain. It is enough work maintaining the site, working on code and once in the blue moon trying to update docs - indeed some of these things get neglected as is.

Also - regarding GIT. AROS has stuck with SVN for a number of reasons, but mainly because using GIT forces it to stay with GIT. Currently, the devs who do work on AROS do so using the scm of choice, and since those all can handle SVN - it makes more sense to use the common denominator. That's why there are public copies of the AROS repository available via git already (e.g. Jasons mirror on github).

Please try to be productive, helpful, and on-topic. Anti-AROS sentiment is not welcome in this thread at all; please start another one if you want to rant in that direction.

Power to the people? Let's see how it goes.

[Kalamatee]

Originally Posted by Locutus

Theres work on a native AROS Git client ongoing, i haven't checked out yet at what state it is now.

It has a long way to go. So far I have it compiled, but some of the code needs fleshed out still (e.g. in posixc), among other things. The binaries run but don't connect to anything yet.

[Kalamatee]

Originally Posted by Toni Wilen

It was interesting and fun (and still is) as long as it was/is not too high level: mainly hardware drivers or dos related (dos is always interesting and weird!), gfx.library, especially intuition and other higher level libs/classes/whatever are very uninteresting and also usually they don't just fit nicely with m68k hardware (like planar bitmaps). I am 100% not interested in optimizing that!

I don't entirely believe you - its all just as much fun if you are able to devote the necessary time/resources, but understandably with the amount you have on your hands with WinUAE low level stuff is easier to fit in ;p

Debugging some big program and finally finding out that problem is caused by (for example) some minor header bug (like wrong register parameter or missing declaration that can only affect m68k), is mostly annoying and really boring.

It is tedious for sure - until you fix some really annoying bug that's been impossible to pinpoint, and suddenly the whole system seems to feel more stable/stuff that you didn't expect starts working also ;D

[wawa]

i would like to mention that the zune compatibility bounty is anyway an ongoing effort, several aros coders have contributed to over a longer amount of time, but none has ever claimed it.
it probably can be considered very advanced if not almost complete, since programs such as odyssey web browser are working on aros.

power2people unlike other bounty hosts, in the wider amiga community, have proven to be very dependable towards developers and the donors.

i think the idea should not be to throw in an insane amount of money at once. rataher to start little and see if such cooperation is mutually satisfactory.