 |
Mardi 11 août 2026 - 07:50 |
| Translate |
|
| Rubriques |
| Réseaux sociaux |
|
| Liste des jeux Amiga |
|
| Trucs et astuces |
|
| Glossaire |
|
| Galeries |
|
| Téléchargement |
|
| Liens |
|
| Partenaires |
|
| A Propos |
|
| Contact |
|
|
|
Today's interview is with Krzysztof Smiechowicz, one of the most active developers on the AROS scene. He talks to us about this operating system and its key applications, such as Odyssey, USB3, Nvidia 3D drivers, EmuV0 and AxRunTime, as well as his future projects and his views on various topics.
Hello Krzysztof.
Could you introduce yourself?Hello everyone. My name is Krzysztof. I'm 45 years old and I live in Poland. I have a degree in software engineering and early in my career I worked as software developer and then moved to different management roles around software developments as well as doing some product management.
What was your
very first computer, and how did you discover the Amiga?My first computer was C64. I saw it first at my cousin's and loved Flimbo's Quest. I bugged my parents until they bought me one - I was probably 10 years old at that time. It was a standard unit with cassette player - no disk drive. I used it mostly to play games with my brother and friends. At some point I started to get tired of games I had and started writing my own games in BASIC. I also dabbled a bit with machine code, but I mostly didn't know what I was doing. While there was a manual covering BASIC and there were program listings in some computer magazines, there was very little information available to be in case of machine code.
A few years later I got invited to another friend who showed me his Amiga 500 and North & South game. It was so amazing. At that moment, I knew I wanted an Amiga. I had to wait some time but I finally got mine Amiga 500. I was probably 13 at that time.
Where does your
nickname "Deadwood" come from?After I got the C64, I started reading computer magazines. One thing that caught my attention was "the demo scene". Since I was already an "experienced" BASIC programmer (in my own, 11-year-old world), I decided scene is the thing for me. And of course if you want to be part of "the scene" you need to have a cool nickname. So I though hard and long about it, trying a few nicknames after I finally settled on 'deadwood' which was a name of a character from a cartoon that I saw on Cartoon Network. Of course by the time I made the final decision on the nickname, my enthusiasm for demo scene somehow evaporated. Eventually, few years later, I did some simple demos on Amiga, but I never joined the demo scene. The nickname stayed on though.
What did you
develop during your early years on the Amiga?My Amiga time was almost entirely filled with games. My Amiga was a standard A500 with 1MB RAM and no hard drive, so you can imagine any productivity work was quite painful with swapping floppies. I didn't use Workbench that much - I didn't really understand how it worked, the libraries, devices, etc. and how advanced it was for it's time. Funny considering I ended up developing system like AROS. The developments I did were mostly done using AMOS. I knew BASIC from C64 so AMOS was na natural step for me. One or two games, some "database" programs for my dad - that's practically it for development during early Amiga years.
Lack of this OS experience in early Amiga years means that even now I'm still discovering sometimes how things should be done "the Amiga way" and I need to be on guard not to introduce "Linux way of working" into AROS. If anything I'd like it to be the other way around (but that is AxRuntime story).
What are
your current AROS/Amiga setups?From everyday usage perspective, I'd say that my Amiga setup is... Linux Mint running AxRuntime. I use Odyssey daily to browse any sites it supports and I use MUI MPlayer to listen to Amiga/C64 radio streams - all while still on Linux.
From development perspective, I have numerous hardware setups, motherboards and peripherals, which I constantly use to test AROS. They are not my daily systems as I constantly swap pieces in and out.
I also have a number of 68k machines, but they are not being used right now. A couple of years ago, I got Amiga 1200 - something that I dreamed as a kid having Amiga 500. I also have a big box Amiga thanks to Timothy Deters sending me one at a quite discounted price.
Testing new 3D drivers on GTX 950 on Krzysztof Smiechowicz's test desk
You're very
involved with AROS. What are your earliest memories of this system? Its launch in 1995? What sparked your
interest in it?I actually joined AROS in 2008, so I was not there for the first decade of system's growth. It was around the time I was in my late 20s. I graduated from university, got a stable job and I had too much free time. I don't recall anymore how I stumbled onto AROS. It was the beginning of retro wave for my generation and AROS reminded my of my happy days with Amiga. Initially, I wasn't doing any system development. I started with working on Murks!IDE, but to get everything working there were changes needed in the system and step by step I moved from application development to system development.
As the retro wave sparked more and more people to get interested in Amiga scene and in AROS we made some great leaps in early 2010s: Poseidon USB stack, 3D drivers for Nvidia cards, Odyssey browser.
Does AROS
development have a roadmap, or is it still "No schedule and rocking!"?It's actually both. In AROS, you can see two distinct efforts that are running in parallel. First one focuses on "cutting edge" topics while the other works more towards "stability". Nick Andrews leads the "cutting edge" efforts over at github.com/aros-development-team/AROS. This is the place where major advancements are developed. SMP, USB3, new hardware support, new AROS architectures - all take place there. Some of these efforts take long time to develop and they can be interrupted in the middle due to real life circumstances. In short, they are not developed "on a schedule". However if you are a technical person willing to experiment and you are not discouraged by occasional instability, I encourage you to try out these nightly builds and support the team with testing.
I lead the "stability" efforts. My focus in on continuity, stability and predictability. Even though these two groups are somewhat separated, new advancements flow into "stability" path, where they get more testing and bugfixing, eventually leading to better outcome for both efforts. Advancements which are not yet stable enough are parked so that they don't cause friction for people who just want to use the system and are not interested in technical details.
As part of predictability, I publish a roadmap, so that distribution maintainers and users know where we are heading and what they should and should not expect. The base of the roadmap is a vision of 64-bit AROS, AxRuntime and pool of 64-bit software that runs on either of those platforms, giving people choice how they want their Amiga-experience to look like. Every few months, I publish a report of work realized so far and next set of short term objectives so that it's visible what progress we are making towards realizing that base vision. The thread I linked above contains history of these reports so everyone can see how our work evolved over time.
You're a key
figure in the AROS world. How many hours a week do you devote to the Amiga/AROS?I'd say that it is around 10-15 hours a week on average. Sometimes, I take a pause for a few weeks, sometimes I hyper focus and spend more time. Most of that time is spend on code integration, testing, bug fixing and helping other users and developers. I myself don't do much raw coding anymore - I think the last thing I developed from scratch was EmuV0. For everything else, there is already somewhere a code that does what is needed and I prefer to port that code instead of writing new code. A good portion of my work is centered around debugging system elements, especially debugging drivers on hardware - and that's always a rather slow process. Every quarter or so I also spend time on synchronizing code between the "cutting edge" and "stability" groups to make sure the divergence between them is minimized. In a way, my work is a glue between system developers and distribution maintainers.
Apart from
yourself, can you name the main developers of the AROS operating system and their roles/work?I already mentioned Nick Andrews. His focus is mostly on hardware side and low level kernel workings, but if needed he also develops user applications or fixes Zune (AROS MUI re-implementation). He is also the maintainer of Wanderer - AROS's Workbench replacements. Neil Cafferkey is mostly focused on network drivers. He recently fixed and migrated most of our network drivers to 64-bit. Bo Stale Kopperud is a fairly recent joiner and is now focusing on getting AROS to work directly on RaspberryPi hardware. In addition, there is a number of people who contribute fixes and improvements on and off to base system.
There is also a few people who don't work directly on core system, but who deserve mentioning. Timothy Deters is supporting AROS developments with hardware and donations. Recently, he has been working on really cutting edge stuff with and around AI on AROS, but that is still in stealth mode. Carlo "AMIGASYSTEM" Spadoni is maintaining leading 32-bit and 64-bit distributions, taking great care for their visual cohesiveness and usability. He also tests and provides feedback on basically any new application or improvement. That's something that is really important to developers. Lastly, Andrzej "retrofaza" Subocz is maintaining AROS Portable distribution and created a specialized, retro-feel AROS.games distributions. He does a lot of games porting, building up AROS software library.
The AROS.games distribution
Can you tell
us about the EmuV0 emulator? How does it work? Doesn't this signal the end of development for 32-bit AROS?As we are migrating towards 64-bit, most of the software we had on 32-bit is open source and can be recompiled. There are some exceptions that are commercial and come without sources and also there is Hollywood, which supports 32-bit AROS but does not plan on supporting 64-bit AROS. EmuV0 is an answer for those issues. We didn't want to leave people behind, especially people who develop and use Hollywood software and thus EmuV0 was created. It's job is to translate calls from 32-bit software and forward them to 64-bit system libraries. It's not a total machine emulation, like UAE, but rather a proxy. It supports just enough to get the needed applications running on 64-bit system.
As for your second question, the vision I described earlier centers around working towards a 64-bit platform. That's something I'm personally focused on and where I put my time. If there is someone interested in maintaining the stable 32-bit version of AROS, I'll happily help to onboard him and support his efforts. The greater the variety of AROS, the better for the system.
The return
of Nick Andrews, another major contributor to AROS, is excellent news. How has the development of USB3
for Poseidon been going? Will these advances in USB 3 also benefit other systems that use Poseidon (MorphOS, AmigaOS)?Nick actually never left, but he sometimes takes longer sabbaticals from AROS and it may seem he is gone. Developments of USB3 are going great. It's actually one the good examples of cutting edge stuff flows towards general user base. Nick did the 95% of work setting up the complete system and testing it in virtual machines. I then took over, tested it on multiple different computer systems, shared with user community for further testing, ironed out bugs and stabilized it to the point it is now in stable beta version.
Trident, the preferences tool for Poseidon
Developments
in modern PC hardware (the phasing out of BIOS in favour of UEFI, the lack of drivers for recent integrated
Intel/AMD graphics chips) are making it difficult to install AROS natively. How can the system overcome
this hardware barrier? Are AxRuntime and hosted mode the only way forward?If you asked me this question six months ago, I'd say that it's not realistic to expect a niche operating system to run on any PC hardware out there. However, today, after I saw what AI agents are capable of and I saw results of experiments done by Timothy Deters with AI agents, I'm not so certain anymore. That's however topic for future.
As for today, I think we have a pretty good offer for potential users. We aim to make the onboarding as simple as possible. First of all if you don't want to buy any hardware, you can run AROS in VM. We support VMWARE, VirtualBox and QEMU. If you have a spare, older PC, you can run AROS Portable (see AROS Portable presentation on YouTube). Again we use there virtualization technology, but make sure you see only AROS and you interact only with AROS. This is the closest for today with regards to AROS running on any PC. If you like AROS enough, there is a list of recommended hardware which you can easily purchase second-hand below 100 euros. That hardware is tested and guaranteed to support AROS.
Lastly, as you mentioned, there is hosted mode and AxRuntime. They too have their fans, while being a mix of Linux and AROS. There's plenty of choice in my opinion.
You are also
the maintainer of the Odyssey web browser for AROS. What are the main challenges involved in maintaining
such software? I imagine the work involved in adapting WebKit is enormous...Yes, the codebase is sometimes... terrifying. Just to give everyone an idea - compiling Odyssey takes four times longer than compiling AROS. That aspect alone is probably the biggest challenge as I often end up re-building the whole project due to how dependencies are structured. If you make a change and then need to wait for an hour to see the result, it can easily suck up all the motivation.
The work of adapting WebKit itself is also challenging. The codebase changes a lot between releases and all the UI and "glue" codes initially developed by Fabien Coeurjoly need to be adapted over and over again. However this has gotten easier since version 3.0 of Odyssey as I switched the base engine to Wayfarer's WebKit. This means that for the part of the work I used to do, I can now use already working codes from Wayfarer. Big thanks go to Jacek Piszczek for making his Wayfarer's codes available to the community.
What features
or optimisations have you added to your version of Odyssey? And what features would you like to add in the future?So far Odyssey is purely in maintenance mode as I'm focusing on AROS migration to 64-bit. This means my main goal is not to lose the heritage of Fabien's work and keep Odyssey usable while updating to newer Wayfarer WebKit versions. Beyond that, when time finally permits I'd like to do two things. The first one would be to do accelerated decoding of multimedia through Nvidia hardware as well as accelerated rendering using 3D. The second one would be to integrated back work being done for AmigaOS 4 into the repository so that there is one codebase for both systems.
Can you tell
us about AxRuntime? How has it progressed since its first release in May 2020? Have any Linux developers
or users shown an interest in it?Has it been so long since first release? I guess it has. There was some interest in the concept of AxRuntime, but so far that interest still came from people already in Amiga community or people using Linux and returning to Amiga community. I released fairly stable updates to AxRuntime in 2022 and then in 2024, adding for example network support so that AROS 64-bit Odyssey can be now used directly under Linux as if it was a Linux application. Any major work however was put on hold. With the roadmap for AROS of migrating to 64-bit, there was little time for AxRuntime. I hope to return to AxRuntime in 2027 though, but now there will be an added benefit of a collection of 64-bit AROS software that can be run under Linux through AxRuntime - something that was not exiting in 2024 (see video about AxRuntime).
Are you planning
to create a modern integrated development environment (IDE) for AROS? (There used to be the Murks! IDE in the past)Indeed there used to be Murks!IDE in the past and I actually started my AROS adventure with working on it. Being young and new to AROS I of course decided that I can re-write it to be "cleaner". Now I'm older, wiser and know better ;).
I personally don't plan on working on an IDE for AROS. AROS developer community is growing, especially since the advent of AI agentic coding earlier this year, so I hope someone else will pick this up. Honestly, with how good the AI tools are right now (middle of 2026), creating a workable IDE should be well within reach of AROS community.
What other
projects do you have in the pipeline for AROS?My biggest project has always been to spark life back into AROS user community, to see it reach and surpass activity levels of early 2010s. All the work that I'm doing with migration to 64-bit, AxRuntime, Odyssey, onboarding developers is all a part of that bigger project. Slowly, all the work being done is starting to bear fruits. Especially this year, as the migration is in its final stages, I can see more and more people joining community, discussing, trying out and sharing experience. There is still more work to be done and I'm sure my plate will be full for next 3-4 years.
What improvements
or features would you like to see added to AROS? And which software would be the most urgent to port or develop?If you think about everyday usage of operating system, most time today is spent in the web browser. You read, work, entertain - all in one tool. For a browser to work close to mainstream on AROS we need three things: good browser engine, lots of RAM and lots of processing power. We have the first - WebKit is open source and Odyssey can use it. Second topic, RAM, is not an issue on 64-bit. We have hundreds of GBs of RAM supported - enough even for heaviest pages. Third topic is processing power. Hardware itself has is - raw GHz, multiple cores and support for latest GPUs. What is missing is ability to use it, so the two improvements I'd like to see is getting the Nvidia drivers to the point where they can speed up browser rendering and enabling multicore support.
Multicore support is actually already developed for 64-bit AROS in "cutting edge" group but is missing testing and stabilization for making it available to general public. On the other hand AxRuntime already has multicore support working and Odyssey running under AxRuntime is visibly faster than same Odyssey running on 64-bit AROS (see video of Odyssey on AxRuntime).
What do you
think of the various current AROS distributions? And what about Paolo Besser's decision to stop working on
Icaros Desktop?They are all great! Each of them takes a different approach to AROS usability and can be interesting to a different group of people. You have one that uses modern visual style and is packed full of software for every occasion. You have one that is small, nibble and just has the bare minimum, letting the user grow it in whatever direction he likes. There is one that focuses on retro style and gamers. I think we cover a good portion of what users can be interested in.
Icaros Desktop will always have a special place in my heart. It was actually the first distribution that really took off. In the 2010s, we really had an image problem. If you looked at AmigaOS 4 or MorphOS you felt that these systems were polished. If you downloaded AROS's nightly build however, it felt like this system is more bare-bone and more for technical inclined people then for regular users. Icaros Desktop changed that - it showed everyone that AROS, even with all its quirks back then, can also be polished. Big thanks go to Paolo for doing that.
Amiberry
Is AROS
for Amiga 68k still being developed? Do you use it, and what do you think of it?In the past, there were developers who dedicated work to AROS 68k - people like Toni Willen or Jason McMullan. Today, there is no such person - it's a position waiting to be filled if anyone is interested. On the other hand, every improvement that happens in AROS, like a fix in a library or new functions added to Wanderer is also next day available for 68k users in form of nightly build. So AROS 68k today grows as AROS grows.
AROS 68k was really one of the key milestones in AROS' history. For a long time AROS had an opinion of being "nice, but incompatible". Then AROS 68k happened and a lot of compatibility fixes were made because it was possible to run 68k software. Eventually we reached a point where AROS 68k was good enough to become ApolloOS and to be used by AmigaKit in commercial products such as A600GS and A1200NG. I think we are in pretty decent place right now with AROS 68k.
How many
AROS users do you estimate there are? How can we increase this number?It's really hard to talk in absolute numbers. We are not tracking system installations and usage, so we don't have any reliable information. What I can say with some degree of confidence is some relative numbers.
First of all I think we have less users than in 2010. Back then, 68k part of Amiga community was stalling. With no improvements in sight, people were looking for alternatives like AmigaOS 4, MorphOS or AROS. Today, situation is quite different. 68k scene turned around with new, powerful accelerators, new AmigaOS updates, games and new software. This means big portion of people went back to 68k from NG systems. I actually see it across the board. AmigaOS and MorphOS forums also slowed and are limited to a few same users compared to situation several years ago.
On top of that, I think AROS community is still relatively smaller than AmigaOS 4 or MorphOS communities. In early 2000s, these two communities created their core users groups. People gravitated to one or the other. What I observe today, when I talk to people, is that they like how AROS is growing, they like the concept and think that it's cool... but they are already invested and accustomed to their choice system and don't have enough incentive to try something new.
With all that said, I think to realistically grow AROS community we have gone outside of current active Amiga users. We can become an interesting choice for people who had Amiga in the past and want to re-live the experience. We are also trying to tap into the wide retro movement - that's what AROS.games distribution is all about. We need to bring people from outside and younger people if we want to flourish.
The donation
system ("bounties") has often been a driving force behind AROS's development. Do you think this model is
still viable today for moving the system forward, or is it showing its limitations?Bounties system indeed worked quite well in the 2010s. We got Poseidon port thanks to it as well as open sourcing Odyssey and many, many other improvements. It worked because the user community was large enough for bounties to grow to sensible numbers. This attracted developers who also saw that there is someone waiting for a given feature and willing to donate for it. Today, I'm afraid the community is not large enough to make this work. Just to give you an idea, with bounties back in the 2010s, we could get 20-30 different people donating. Last year, we tried to organize a fundraising for a quite important topic and ended up with 5 people willing to donate.
The approach is still valid, but it will have to wait for community to restore its strength. Also, right now, everyone has more opportunities to support selected developers for their already done work. PayPal and platforms like Ko-Fi or Patreon are perfect ways to support developers. I'd encourage everyone who can to donate - developers really appreciate this not only for monetary value but also as a sign that their work is important and valuable to someone.
What are
AROS's current strengths and weaknesses compared to MorphOS and AmigaOS 4? Are you also interested in
contributing to the other systems? Have you received any requests from those teams?I already talked a bit about where MorphOS and AmigaOS 4 got things much better than AROS. They both got in front of AROS because they offered a polished and modern experience when the 68k looked like it is going to end. On top of that, the architecture choice of PowerPC allowed for transparent running of 68k application which made people feel right at home with those two systems. This made them natural choice for users migrating from 68k and both MorphOS and AmigaOS 4 still today receive the "dividend" from that.
If you look at AROS today, I'd name two main strengths. First and obvious one is cheap, modern and powerful hardware. You can't really beat price/performance ratio for x86_64 machines with anything else. Believe it or not, but I know a person who runs AROS on a server machine with 64 CPU cores and 256 GB RAM. If you are interested, you can buy one of the recommended boards or just get AROS Portable and run it everywhere. This low threshold of entry is something I consider a strength when I think about people from outside current Amiga community.
Second is about AROS openness. The source code is out there in public, you don't need to sign any NDA or legal documents, you can just clone the repository and start taking it in whichever direction you want to take. What's more, that code is not going to be suddenly pulled down by some corporate decision. As long, you respect the licence, you can build your project on top of AROS and know that your work won't be cancelled at someone's whim. This openness was always attracting outside system developers to join AROS and I think it makes us ideally positioned for the new world of AI assisted software development. Over the last couple of months, we actually see an explosion of people taking AROS as a based and doing their "dream" projects with it. You can't do that with any other operating system in our community.
What is your
view on artificial intelligence in the field of software development? Do you use it for your Amiga/AROS projects?Yes, I use AI tooling in my development work. Since I do a lot of porting and integration, AI tools help me a lot with understanding new codebase, tracing code execution or speculating on root causes of bugs. They are amazing at answering questions about code - something that I could spend 10-20 minutes searching through code and still miss cases, AI does in a minute and is spot on most of the time.
Taking a bit wider view on how AI assisted software development can impact Amiga community, I'd say there is now a real opportunity for AmigaOS 4, MorphOS or AROS to reach usability level of Linux or better in less then a year. I've seen what Timothy Deters was able to achieve in days with his AI workflows. It takes a right person and right approach but think about it - all the productivity software you will ever need has already been written and is open source. What you need is a massive porting effort and AI is damn good at porting. It might not create the highest quality engineering work, but it will iterate and iterate until job is done. It never sleeps, it never rests.
One may of course argue if a system has all major software ported from Linux is still going to be different than Linux. Here my answer is again AI. UI coding is repetitive, AI can do repetitive work and make those programs use native Amiga system UIs. We are already at place where you can take a frontier AI model, tell it "Re-create GIMP UI in MUI", go to sleep and in the morning you will have first working version.
What do
you think of Christian Simpson's takeover of Commodore and the easing of tensions between Amiga Inc.
and Hyperion? Does this have any tangible impact on your day-to-day life as a developer?I'm not really following what is happening with the new Commodore so can't answer that. As for Amiga Inc. and Hyperion I do loosely follow latest news but I never had contact with either of those parties and I doubt the resolution of their disagreement will have material impact on AROS.
Is there a
question I haven't asked that you'd like to answer? Or an anecdote you'd like to share?No, I think I already covered a lot with my answers, sometimes more than the question asked for ;)
Any final
message for the Amiga/AROS community?I'm always amazed how Amiga community continued to thrive after the fall of Commodore. People always found ways to overcome obstacles, create new hardware and develop needed software. People from different parts of the world helped each other to make Amiga experience better. Keep this sprit up and encourage younger people, your kids, your students, your neighbours to get interested in Amiga to keep the Amiga dream alive.
|
|