We’re all aware of Haiku, the open source re-implementation of BeOS. This week, I found out a similar effort is underway to recreate OS/2 Warp 4 as an open source operating system.
osFree is a Free Open Source software operating system development project, aiming to replace eventually all OS/2 subsystems with Open source analogues. It aims for OS/2 Warp 4 (Merlin) as a base compatibility system, which does not mean that we will not support features of newer (OS/2 Warp Server for e-business, eComStation and ArcaOS) OS/2 versions. This includes rewriting not only user-level code but the OS/2 kernel too.
The project is very much in its early alpha stages, so don’t expect to boot into osFree any time soon. That being said, they have made progress over the years, and their current status explains how far along they are.
osFree project is deep in the alpha stage. At the present time the osFree project work with low-level parts of operating system like loader and related tools. We also try to develop an experimental prototype of OS/2 personality for L4 microkernel. Also we have set of command line tools like CMD.EXE and file/disk maintenance utilities.
I absolutely adore OS/2, and while I understand all too well a project like this won’t be finished overnight – look at Haiku – I do love that it’s being worked on.
OSFree has been going for a while now. As far as i’m aware, it’s nigh on abandoned now, and has been for many years. Last i checked there was a sole person commiting changes around once a week.
I believe it’s based on the L4 kernel, and has some semblance of OS/2 compatibility, but there has never been any boot media available so i never tried it out. I think there’s a few useful open-source OS/2 components used in it, or will be used, but it’s so early in development, i don’t think much is of any use.
Neat idea, just lacks developers. It’s also reassuring to know there’s projects that are more incomplete than ReactOS.
Last commit is from October… it’s dead all right. When I saw the article headline, I was like, is it alive again? But that was false alarm.
Last commit was today. You are probably talking about osFree userland commits, which were in October. But the project consists of several git repos. Currently the OS/2 personality repo undergoing more work than the Userland repo. And yes, a half year pause indeed is not much. The project develops very slowly. It is sleeping sometimes, but once it becomes resumed again. Unfortunately, the project lacks manpower. There are 1-2 people active, But they also busy with the main job, hence less attention to the project. Unfortunalely, a there are a little people who want’s help us, so development is very slow. But we don’t give up.
osFree exists for a long time. It is not abandoned, it just lacks manpower. There’s only 1-2 developers active currently, including me. It’s being worked on, though, I have many projects, besides osFree. I need to pay attention to each of them. While paying attention to one project, I need to pause others. Unfortunately, manpower is limited. Also, I have main job, so osFree needs to be a hobby project. But I try to pay more attention to it, despite problems. Also you can help with code or donations. Yes, it is based on some L4 kernels. Currently, L4/Fiasco. Also, port to Fiasco.OC/l4re and Genode OS framework are in progress. Regarding the boot media, there is a boot CD in the download area. It was updated a while ago. Yes, it lacks developer. We have 1-2 active developers vs 10-20 with ReactOS, so don’t expect quick progress. But you can help us if you like it.
At least OS/2 Warp 4 is a fixed target, theoretically making the task more achievable than ReactOS.
ReactOS has a fixed target… Windows Server 2003, an that isn’t changing what they are doing is occasionally adding APIs for later versions as it is convenient. Still the hardware driver support and kernel target itself is the same.
Since ReactOS uses a LOT of Wine code, application compatibility is nigh on par with Wine. There’s good chunks of code in Wine that need rewriting for ReactOS, opposed to Linux , but generally much of the application compatibility is down to the Wine project.
The kernel and general userland however, is still based on NT 5.2
Yes, OS/2 API is small and lean, compared to the Windows API. So, we have lesser amount of work to do, compared to ReactOS. So, the work is achievable with lesser manpower. Though, ReactOS has WINE codebase which they are using. We haven’t a project which code we can directly borrow like ReactOS from WINE. Though, some windowing functions in OS/2 PM are similar with some WinAPI functions. So, in theory some code could be borrowed from ReactOS/WINE. Windows API is huge, it contains thousands of functions. OS/2 API is smaller. Base kernel API consists of several hundreds of functions. A full windowing API contains about 800 functions only. BTW, ReactOS’ success was an inspiration for us. Originally, we wanted to use ReactOS kernel as a base, but after some research, we choose L4 family kernels (originally, L4Ka::Pistachio, now L4/Fiasco and Fiasco.OC). We’re fans of true microkernels. So, our dream is a functional equivalent of IBM Workplace OS, based on newer microkernel technologies, like L4-based kernels. They are far more advanced than older Mach kernels, on which IBM Workplace OS, NextStep, MacOS X, GNU HURD, and Win NT 3.0 were based. Currently, we think that the most advanced L4-based project is Genode OS framework (https://genode.org). We’re big fans of it. It has a potential of becoming the most popular community-based microkernel project. We currently creating a port of osFree OS/2 personality to Genode (OS/2 personality is currently based on L4/Fiasco and L4Env, which are outdated now and replaced with Fiasco.OC and L4Re framework. So, it is currently being ported to L4Re and Genode). Unfortunalely, the progress is very slow with current manpower and the amount of free time.
How can I throw monwy at this project to get it going? I’d like to throw money at ReactOS also, as I’m not a developer, but would like to help where I can. I’m not talking $1.00 I’m talking +$500 at each as I’m just one person. LOL
Save your money, throw it at ReactOS. ReactOS is a much more stable and complete system at this time than OSFree.
They also have donation pages. https://reactos.org/donate/
Stable and reactos don’t’ belong in the same sentence… further along yes but not stable. I realize you qualified it relative to osFree.. but still neither is day to dayable.
You could dogfood Haiku OS in some cases… it’s missing some modern stuff but you could get by mostly without people thinking you had died.
Several people asked us about how they could donate to the project. Originally, we had a donation page at sourceforge.net, but sourceforge.net closed its donation service. Now we created a donation page at http://osfree.org/doku/en:donate. Also, we have a Patreon project https://patreon.com/osfree/ for those who wants to be our patron on monthly basis. We have a number of side projects, except osFree itself, like VirtualBox port to OS/2, VirtualBox OS/2 additions (http://trac.netlabs.org/vbox/), universal FAT/FAT32/exFAT driver for OS/2 (http://trac.netlabs.org/fat32/), etc. Here, we will show our progress on these projects, and get support. Also, we have a bitcoin wallet for one-time donations. So, you’re are welcome. Also, you can help with system administration, web-development of the web site, system development and design. See here http://osfree.org/doku/en:develop. Of course, ReactOS is more promising project than ours, but you decide.
For a more practical approach, just fire up a virtual machine…
But sure this project could bring a lot of fun for interested devs.
There is a patreon setup as of last year, https://www.patreon.com/osfree apparently they also maintain the Vbox extensions for OS/2.
I can’t begin to comprehend why anyone would trivialize independent OS projects, discourage supporting them, or find it ‘reassuring to know there’s projects that are more incomplete than ReactOS’.
This is the sort of swill that discourages OS Developers and support for projects that have real veracity and progress like osFree and only wedges the world harder and faster into the Micosoftlinuxalpahbet hegemony that the powers that shouldn’t be regard as both desirable and inevitable.
Is this what you really want to support? I hope some will more carefully weigh and consider their criticism, who they reward, and who they punish through their words, action and inaction…
Eh it’s just being realistic… on the other hand I imagine most people here are quite excited to see the progress made on each project. And I’ve even donated to some in the past.
Imagine if R Stallman (GCC), L Torvalds (Linux), B Paul (Mesa), J&L Jolitz (*BSD) & J Hall (FreeDOS) & B Amstraad & E Youngdale (Wine) took a “Why bother” attitude. When you consider that the foundation for almost *ALL* of modern IT is built upon these enthusiasts’ pet projects even when established, entrenched & superior products were already widely available)
Remember when Linus wrote (what was tp become linux) was just for learning about the 80386’s hardware capabilities. It wasn’t going to be professional like Minix, IIRC.
Who knows,, in 20 years we might all be running a derivative of osFree. Some would argue that starting with L4 makes it fundamrnally more sound than Linux. Perhaps they coiuld piggyback off the work done by Google on Fuschia. The biggest problem with OS development is device drivers. Having usermode drivers with a simple kernel API could make it easier for this new wave of microkernel-based OSes to share a common driver base.
pjlahaie,
That’s all well and good, but re-reading cb88’s posts doesn’t make me feel he said anything particularly disparaging or untrue. I actually think I’m more pessimistic than he is.
I think that’s a great idea. Most indy operating systems simply don’t have the manpower to write drivers to support everyone’s hardware, but if they could somehow team up and join forces, it theoretically might be solvable. In actuality though I think the fruit of such efforts would most likely take the shape of linux kernel drivers because that’s the most comprehensive source of drivers available. Yet, to be honest, I’m not particularly fond of the state of linux drivers. They’re not stable (by which I mean well defined interfaces). They don’t handle exceptions/cancellation events well (a flaky network/usb connection can create unkillable processes). They’re not robust against low memory conditions. They’re not suitable for RTOS, They’re not asynchronous. And on top of technical issues the GPL license could well be a sticking point when linking into other kernels (at least compared to more permissive BSD-style licenses).
IMHO there’s a lot of room for innovation and improvement. So theoretically there are new opportunities for indy projects to do better. However this is where my pessimism comes in. For all it’s weaknesses, linux has critical mass. I think you’d agree this is the biggest challenge for new indy kernels. Even if they are built on superior foundations, they may not receive sufficient attention to make a dent in the driver problem. What’s the likelihood of nvidia and others contributing drivers to a project with 0% market share?
Indy startups don’t have the same opportunities to shape the world like they did when the industry was young. It depresses me that indy operating systems today make up a combined 0%. Don’t get me wrong, I appreciate your positive attitude and I hope you keep it up. Seriously it would be great if you keep providing alternatives to the pessimistic visions of the present & future, haha. The world could use more optimism 🙂
Yes, the device drivers problem is the most importaint. We cannot write drivers ourself for each and every device. But as your said, there are Linux drivers. And not only the Linux ones. The OS frameworks like L4Env, L4Re, Genode, and microkernel projects like Minix3, have DDE’s (device driver environments) — libraries which enable some drivers from different OS’es to be compiled and reused on another OS. E.g., Genode has DDE’s for Linux drivers. Actually, they enable porting drivers for USB, Intel NIC’s, wifi stack, Intel GPU. Yes, Linux has a big codebase for drivers, but it has many disadvantages, like unstable kernel API’s for drivers, and others, as you mentionsed above. So, there are DDE’s for OS’es other thna Linux. E.g., Genode has DDE’s for OpenBSD audio drivers, IPXE NIC drivers, NetBSD rump kernel port for NetBSD file systems, also, there is a small DDE for drivers from Fuchsia OS/Zircon kernel, like PS/2 driver. One of the reasons we chose OS frameworks like L4Env/L4Re/Genode were these DDE’s that were available on them. Also, we see that code reuse from other OS’es help our OS/2 system to survive. We have a port of ALSA for sound drivers (called Uniaudio), a port of Linux AHCI driver (called os2ahci), port of Linux NIC drivers called multimac, a wrapper for windows wifi/NIC drivers called GenMAC etc. Of course, huge drivers like Nvidia driver, are a big problem. Probably, open source Linux driver could be reused. Genode reused Intel GPU driver, for example. Most OS’es (like OS/2, Haiku) use VESA drivers for 2D, often with no accelerations and software OpenGL. Often, this is sufficient.