forum GNU Linux PowerPC Notebook

Software => Others OS => Topic started by: Casper on April 30, 2016, 04:32:52 PM

Title: Redox OS
Post by: Casper on April 30, 2016, 04:32:52 PM
This might be an interesting idea for an OS.

Basically, they build a microkernel in the Rust programming language with a focus on safety and speed.
The idea is to build a modern Unix-like OS that sheds many of the legacy stuff that has built up on Linux.
The project is just over a year old but seems to be a fast moving target with over 40 developers that contributed already.
The only problem is that they are only developping on IA-32, AMD64. If we would be able to work with them to develop a ppc64 version, we could have a modern, fast and interesting new OS.

Title: Re: Redox OS
Post by: carguyty on May 07, 2016, 01:28:52 AM
I saw this and thought it was very interesting. So I posted a reply to start a dialog. The reply was a small essay and now it isn't showing up. Before I muster the strength to try to write something similar, was my previous reply deleted?
Title: Re: Redox OS
Post by: Casper on May 07, 2016, 11:06:26 AM
I don't think so. Sometimes the website is overloaded, so it probably was just an error. But I'm interested in reading what you have to say.
Title: Re: Redox OS
Post by: Carlos on May 07, 2016, 01:12:39 PM
carguyty, as you could see on your post count, this was the first post you wrote. So I assume it was an error.
I would like also to know what you have to tell.

Edit: And as far as I know, we never deleted posts. We modify posts.
Title: Re: Redox OS
Post by: carguyty on May 15, 2016, 03:52:30 AM
Cool! I'll try to remember the spirit of my reply....hang on because this is LONG. And it's long because I really care about this mission and I want to start a real dialog.

It's very important to remember that we are trying to invade a very well established market paradigm for computing. PowerPC is only strong (and thankfully gaining traction) in the high power enterprise market. I don't think a notebook platform fits the big data number crunching arena so lets say that we are reaching out to power users and workstation enterprise. We know that this is going to be an expensive offering compared to the available offerings from other established companies. We will simply not have the economies of scale to be competitive directly with the brands like Dell and HP. For the sake of considering any operating system, I believe that we must have a history lesson.

Way back in the earlier days of PowerPC, the AIM alliance was able to leverage the reputation of one company against the production ability of another against the strong engineering of the third, and they all shared the minimal risk (pun intended) left over.

Apple did their thing, and endured a very turbulent 10 years with the beautiful architecture. Some, however, think that it was always meant to be a bridge to help them get into the X86 architecture. As much advancement as had come out of the AIM, there wasn't anything really revolutionary beyond the things that IBM added to the ISA (DDR, ALTIVEC, etc...). With all their money and marketing, they never got much in the way of traction due to the difficulty developers had in porting existing products to a unique hardware/software protocol.

BeOS took a liking to the PowerPC architecture way early on. It was so simple and beautiful, that his team sat down and wrote a new OS from scratch in a way that took advantage of endieness (sp?) and memory management and all the things that made PPC unique. And it was FAST! Fast as lightening compared to ANY other OS of the day (even OS 8 and 9 couldn't compete because of how much 68K emulation was still buried in the code). BeOS worked on modern principles with a modern language and unburdened itself of all the legacy paradigms of the day that were dragging down Windows and UNIX. It was small, light and hella fresh! But there were no developers! Nobody willing to take a chance on a budding operating system for a quirky chip from a company that didn't exist 2 years prior. There's no way to guarantee to first tier devs that their product would ship in any quantity. No third tier devs could afford to take the time to learn the language and syntax needed to make new apps. There was just no convincing the market that we needed this fantastic combination.

A little bit better established but still mired in legal headache, Amiga OS was unable to move to PowerPC effectively even thought they had an outstanding JIT emulator that allowed them to carry their legacy software to a potential fountain of youth. While their niche was in the video market and seemingly removed from the horsepower wars of the mainstream chip manufacturers, they have been all but abandoned by a small hobby community that must make due with modified network equipment and layovers from the PAPR and CHRP era. A-Eon has announced a new "modern" board that will leverage the 64-bit PPC chip as well as DDR3 and PCI-e as well as many other items that will allow them to nurse along their legacy software as well as provide virtually uncharted and unfettered access to the current status quo of computing. But they will always rely on bit players who have extra time to work on bug reports and modify open source ports various utilities and classic software run in low level emulation. This may never be anything more than a large scale science project for a few people that remember the hope Commodore brought to market so many years ago.

So here we are with Redox. I've read all I can find. I love it! It's light and fast and amazing. It's possible to port the language and the OS as easily as any other. Documentation is rather basic right now, but the openness of the Rust means that working out proper compiling should be straight forward. Redox is bucking the standards Linux (which has long been lambasted for its way too diverse options used to support legacy hardware and software) has long held in esteem. Even their license is looking more honest and open than the current GPL. I agree that this is all very much aligned with the perceived mission statement of Something like this COULD certainly help show off what a powerful tool PPC has become. But what risk would a developer need to take on to enter a super ultra mega niche like this. Are we in the business of putting PowerPC back into the hands of the masses? Would this be an advantageous move for both short and long term? How quickly could we get major players onto our platform in order to show the majority that we are both a company that is serious about returning our investors money as well as a gateway for supreme hardware to land on the desks of people trying to do actual work?

These are some of the questions and the arguments I have with people when I try to explain why I still use a PowerPC computer in this age. I want this architecture to be mainstream again. With the open ISA from IBM and the many companies (TYAN, Suzhou PowerCore, Google, Facebook, etc...) that have worked to develop it all took to ditching AIX and favored Linux, the people who invented QuickTransit took notice and worked around a translation layer for X86 apps to have an easier time porting to PPC-64. And it's helping. I don't think Linux is the long term answer (I personally love the classic Macintosh Toolbox and I think that it could be modernized in every way while still maintaining its speed) but it may be the only option with enough support for us to get this product on as many desks as needed to bring down unit costs enough for mainstream consumption to be in our future.

Title: Re: Redox OS
Post by: Casper on May 16, 2016, 08:58:21 PM
Of course Redox will need to quickly gain some app support if they want to maintain their momentum. But imagine if they could have a full production suite all programmed in Rust.

Personally, I don't think MIT is a better license then GPL, but I do hope their program gains some ground, if not to convince Linux to trim off some fat.

It might be a good idea to contact them and ask for the possibility of a PPC port.
Title: Re: Redox OS
Post by: carguyty on May 18, 2016, 06:26:58 AM
If I recall, it's the compiler that needs to get nailed down. The compiler for X86 is what kept it from getting cut out of the supply chain. The compiler was well documented and available to all the developers. The compiler wasn't as ready when Intel tried to release the Itanium and it hampered the plans they had to kill X86. The issue is translating Rust to PPC assembly.

I like the idea of working with them for a future release of a PPC native Redox. Porting the OS is one thing, but porting an existing suite of core apps is going to be something else. Fortunately there are wrappers that could help. It sounds like the OS is crazy light and some built in JIT emulation wouldn't kill the speed. Perhaps work on something like a Java VM could be the stopgap measure we need to get a useable desktop. It would surely help building a proper web browser. I'm learning Java right now. Moving to Rust isn't the best option for me academically, but I would love to collaborate as a side project.

Do you have any interest in approaching the core team with me?
Title: Re: Redox OS
Post by: Casper on May 18, 2016, 09:19:37 AM
Sounds like a good idea. We can use all the OS support that we can get.
Title: Re: Redox OS
Post by: carguyty on May 19, 2016, 04:45:33 AM
Sounds like a good idea. We can use all the OS support that we can get.

That's what drives a hardware business. Being accepted by the masses, no matter how they want to use the technology. ;D