OSDev.org

Hi,

I've checked the Notable Projects page on the wiki, and I feel the need for some moderation. It's starting to turn into another "OS projects" page. Here's what I think:

MikeOS, DawnOS, Pedigree - someone has removed these, wonder why. They are more mature than many on the current list. DawnOS for example is still actively developed, as well as it's SUBLEQ emulator. Pedigree might not be developed any more, but it is still one of the greatest achievement of one of our community members (ipv4/6 networking, GNU userland compatible etc.).

9front - this is a fork, should not be listed as an OS development project (it is not written from scratch, it's a fork). We do not list all the BSDs or Linux either, right?

FreeDOS, Haiku, SerenityOS, Visopsys, Kolibri OS, LK - are these written by this community? I don't think so. We don't list Oberon or Minix either. Otherwise okay, specially if at least one of their developers are members on this forum.

qword - this should be deleted. It is not very mature, no userland at all, no author given, etc. This clearly belongs to the "OS projects" page, not here.

Sortix, ToAruOS - these are ok, just missing author's forum name. Sortix and klange should be mentioned by name I think.

I think we should make some guidelines on what fits on the "Notable projects" page. Here's my two cent:
1. written by an OSdev.org community member
2. should be mature enough, run on many VMs and/or real machines (means relatively decent driver support and relatively bug-free kernel)
3. has a working shell / fully usable GUI with some userland applications (just a kernel does not qualify, it has to be an OS with at least a few apps)
4. being self-hosting is a strong point for inclusion on the list, but not a must have
5. network capability is also a strong point for inclusion on the list, but not a must have
6. ...

Thoughts?
bzt

That all seems reasonable to me.

bzt wrote:MikeOS, DawnOS, Pedigree - someone has removed these, wonder why. They are more mature than many on the current list. DawnOS for example is still actively developed, as well as it's SUBLEQ emulator. Pedigree might not be developed any more, but it is still one of the greatest achievement of one of our community members (ipv4/6 networking, GNU userland compatible etc.).

I think being actively developed was originally one of the goals of this list vs. constantly cleaning up the "OS Projects" list of dead things, though some projects are 'established' enough that they should probably be grandfathered in and remain here. That said, DawnOS does not belong here if not just because its author has engaged in cyber attacks against the wiki and forum and I'd prefer we don't give them any free advertising.

bzt wrote:9front - this is a fork, should not be listed as an OS development project (it is not written from scratch, it's a fork). We do not list all the BSDs or Linux either, right?

FreeDOS, Haiku, SerenityOS, Visopsys, Kolibri OS, LK - are these written by this community? I don't think so. We don't list Oberon or Minix either. Otherwise okay, specially if at least one of their developers are members on this forum.

In my opinion, the list should not be restricted to projects developed by this community, though I do agree that a line may be needed between "notable (hobbyist) OS projects" and "huge OS projects with decades of consistent development and dedicated resources". Serenity is fine for inclusion, Visopsys remains a classic, Haiku is probably beyond our scope, and LK is a toss-up due to its usage by Google.

bzt wrote:qword - this should be deleted. It is not very mature, no userland at all, no author given, etc. This clearly belongs to the "OS projects" page, not here.

qword has a GCC port and its developers are maintaining their own ACPI/AML library, so I think it's at least reasonably 'mature'.

bzt wrote:Sortix, ToAruOS - these are ok, just missing author's forum name. Sortix and klange should be mentioned by name I think.

I don't see why mentioning the author's usernames is necessary, and both are discoverable from the provided source/website links pretty quickly.

bzt wrote:I think we should make some guidelines on what fits on the "Notable projects" page. Here's my two cent:
1. written by an OSdev.org community member
2. should be mature enough, run on many VMs and/or real machines (means relatively decent driver support and relatively bug-free kernel)
3. has a working shell / fully usable GUI with some userland applications (just a kernel does not qualify, it has to be an OS with at least a few apps)
4. being self-hosting is a strong point for inclusion on the list, but not a must have
5. network capability is also a strong point for inclusion on the list, but not a must have
6. ...

Thoughts?
bzt

We can debate the particular factors for inclusion, and I think all but #1 here are reasonable - I don't think the list needs to be restricted to projects by people in this community.

klange wrote: We can debate the particular factors for inclusion, and I think all but #1 here are reasonable - I don't think the list needs to be restricted to projects by people in this community.

I think it is helpful that the notable OS pages show interesting/unusual/advanced projects where the developers can be contacted via this forum. It makes the resource more useful.

There are plenty of pages on the internet covering notable OS projects, but few allow much personal interaction with the developers.

We could have several lists (but I really care only for (1 a)) :
1.) hobbists' projects
a) full-featured, have at least some applications
b) in the making, not yet stable
2.) mainstream OS like BSD, Linux, MacOS
3.) historical OS: you know like VMS, CP/M, Unix 6
4.) open source or PD clones of proprietary OSs: FreeDOS, ReactOS, Haiku, etc.
5.) OSs from universities: Oberon, Minix

You probably can guess that some lists would be large, for example (1b).

bloodline wrote:I think it is helpful that the notable OS pages show interesting/unusual/advanced projects where the developers can be contacted via this forum. It makes the resource more useful.

There are plenty of pages on the internet covering notable OS projects, but few allow much personal interaction with the developers.

I second that.

bzt wrote:MikeOS, DawnOS, Pedigree - someone has removed these, wonder why. They are more mature than many on the current list. DawnOS for example is still actively developed, as well as it's SUBLEQ emulator. Pedigree might not be developed any more, but it is still one of the greatest achievement of one of our community members (ipv4/6 networking, GNU userland compatible etc.)

I agree MikeOS shouldn't have gone. A spot of inappropriate snobbery there, I think. Or possibly jealousy: Just because it seems easier to make a 16-bit BIOS-based OS than to hammer a gcc/elf/what-have-you monster into shape doesn't mean MikeOS isn't notable in its way. I can understand such jealousy, but still...

I don't remember Pedigree specifically, but I am opposed to removing OS projects from notability simply because they're not currently maintained. There are some very interesting concepts in some unmaintained OSs, including some which haven't been maintained for literally decades, and they still run with suitable emulation. ... Perhaps they could go in the "interesting OS" page I came up with below.

klange wrote:That said, DawnOS does not belong here if not just because its author has engaged in cyber attacks against the wiki and forum and I'd prefer we don't give them any free advertising.

I'm very sad to hear that. Rational or otherwise, I rather liked the project, but after hearing about this... well, I think its author needs a long time to cool down... and maybe a hug and someone to tell them they are worth something outside of their achievements.

bzt wrote:9front - this is a fork, should not be listed as an OS development project (it is not written from scratch, it's a fork). We do not list all the BSDs or Linux either, right?

Uh... I was initially surprised at its inclusion for the same reason you are, but on the other hand it's a smaller project than some of the POSIX systems forum members have developed. It was never used on anything like the same scale as BSDs or Linux, but again I get the feeling it has had more use than the top 5 member projects put together. (Managarm might be an exception?) It runs a few websites & mailing lists. A college teacher had it on a 64-core system and taught students with it. Its parent OS has been through the Himalayas with the file server on one Jeep and terminals/compute nodes on others! They -- 9front and Plan 9 both -- are more mature than most projects here, but they're also more manageable and customizable than most other OSs, especially if you're a good C programmer. So really, I have no idea whether it should stay or go, but wouldn't be surprised if it goes.

bzt wrote:FreeDOS, Haiku, SerenityOS, Visopsys, Kolibri OS, LK - are these written by this community? I don't think so. We don't list Oberon or Minix either. Otherwise okay, specially if at least one of their developers are members on this forum.

Visopsys is a one-guy project with maybe one or two interested contributors. I feel it belongs here in spirit. I'm not sure this counts: its killer app is partition management; it's useful to us. FreeDOS Kolibri OS might be a bit too big and popular? FreeDOS's development history looks like Linux's: very many people have contributed components. Kolibri has quite a lot of contributors now. On that point, 9front is slowly gathering more contributors. The others I don't know at all, except Haiku seems to be in the same class as FreeDOS and probably has many more contributors than Kolibri OS by now.

I'd like to suggest a new page for research operating systems and others with interesting ideas. I mean Oberon, Plan 9, maybe Linux. I'm sure there are others. Actually, speaking of ideas, I like to refer to Kolibri's very simple GUI API sometimes. I'm not sure where to draw the line on what would go into such an "interesting OS" page.

bzt wrote:I think we should make some guidelines on what fits on the "Notable projects" page. Here's my two cent:
1. written by an OSdev.org community member
2. should be mature enough, run on many VMs and/or real machines (means relatively decent driver support and relatively bug-free kernel)
3. has a working shell / fully usable GUI with some userland applications (just a kernel does not qualify, it has to be an OS with at least a few apps)
4. being self-hosting is a strong point for inclusion on the list, but not a must have
5. network capability is also a strong point for inclusion on the list, but not a must have

1: Again, I think Visopsys belongs here in spirit, but that's just a feeling really.

I'm lumping 3 and 5 together and saying I think an OS should be required to meet these only as apropriate for its goals. A edicated router/firewall OS would very strongly meet point 5, but by design wouldn't meet point 3. Likewise, a little OS which can't network but works well and has nice apps and games would meet point 3 very well, but not 5. I'd say an OS should meet at least one of these rules for inclusion.

2: uh... I mostly agree, but there's a case to be made for KVM-only OSs for VPS use. I guess if the rules are a bit flexible and such an OS meets other points, this is a good rule. The "apps" in this case may be server programs, but we don't need to specially note that, do we?

4: Self-hosting is such a strange thing. Some Forth designs can be self-hosting with less code and effort than it takes to reach the stage of loading user programs at all in a system built with gcc & binutils. Those primitive Forths wouldn't have a filesystem and would load source rather than binaries, so how much is the self-hosting part worth? All the same, self-hosting is a clear milestone in the development of more complex OS, so it's a useful guideline for including them.

@PeterX: I feel that's too many divisions. For example, some OSs which weren't developed in universities had very interesting ideas.

Okay, you all made valid points why include one-man projects for non-members, and I agree. But then we should mention TempleOS. Funny or not, you simply cannot miss that in OSDev

eekee wrote:I'd like to suggest a new page for research operating systems and others with interesting ideas. I mean Oberon, Plan 9, maybe Linux.

I second that! I also have a feeling that some OS on this list should go there. Maybe not "research" but "more advanced" or "done by larger teams"? I think of remakes like FreeDOS, Haiku too. Let's just call this "List B" for now.

So to summarize your posts (correct me if I miss something):
MikeOS, Pedigree - should be restored
DawnOS - ?
9front, FreeDOS, Haiku, LK - should go to List B
Kolibri OS - ? (either leave it here or move to List B)
Sortix, ToAruOS - add forum member's name for easier contact
qword - remove for now

About qword, I still vote for remove. The link goes to a repo that has nothing but a minimal kernel, there are no userland repos (except for an unfinished libc), not even a shell, and it's homepage is dead for more than half a year now (just reads "Server being setup come back later.") I honestly don't know why do you think it's self-hosting. A screenshot with gcc is not enough, because it has enormous amounts of build dependencies (python, pip, meson, nasm, autoconf, bison, etc. etc. etc). Don't get me wrong, qword looks interesting, and it should be listed under "OS Projects", it's just not ready to be listed in "Notable Projects" IMHO. Several OSes on the "OS Projects" list are more advanced and would deserve to be called "notable" more (in my humble opinion).

Cheers,
bzt

bzt wrote: But then we should mention TempleOS.

Well, the name of the list is Notable projects, and it is hard to be more notable in OSDev than getting multiple internet communities interested in you, including a Linus Tech Tips video on YouTube.

LOL we should mention TempleOS! Anyone remember when it was called Losethos?

I just thought of something to moderate what I last posted because I forgot rdos. rdos (OS) gets some pretty serious use, now including handling GHz-rate sample data from semi-custom hardware. Excluding rdos from notable projects would be ridiculous, so I'm regretting using examples of the use an OS gets as a point against it.

eekee wrote:LOL we should mention TempleOS! Anyone remember when it was called Losethos?

I just thought of something to moderate what I last posted because I forgot rdos. rdos (OS) gets some pretty serious use, now including handling GHz-rate sample data from semi-custom hardware. Excluding rdos from notable projects would be ridiculous, so I'm regretting using examples of the use an OS gets as a point against it.

It probably should be there. It has a kernel with USB drivers for UHCI, OHCI, EHCI, XHCI, and USB 2 hubs, IDE and AHCI for discs, RTL8139 and RTL8169 for networks, and HD audio. It boots with EFI using GPT and BIOS using BIOS disc formats and supports FAT12, FAT16, and FAT32 filesystems. It includes support for both older CPUs with PIC & PIT and newer with APIC which also includes support for multicore operation. It has a full libc in the OpenWatcom project, which also allows building kernel drivers, and userland applications, libraries, and DLLs. It has a command-line originally ported from FreeDos, and a GUI the includes widgets for labels, buttons, list boxes, and images (supporting PNG, JPEG, and BMP formats). As for debugging, it supports remote debugging over TCP/IP from Windows using OpenWatcoms debugger that also can trace into the kernel, a kernel debugger, and a crash debugger.

It's not self-hosted, but the OpenWatcom tools like the compiler, assembler, linker, and debugger can be built for RDOS, and they can be launched from the command line. The only thing that remains is to port the build environment itself, but this has low priority since it is more convenient to develop on Windows and use remote debugging.

It runs on several thousand commercial installations as a controller and as a payment terminal.

In the more private domain, it's used as a home controller for my heating system which communicates with a set of my own PCBs that measure temperature and control a termomotor connected to a radiator. Another PCB contains relays that turn on and off circulation pumps. These PCBs are connected on a custom 485-based bus to a central PCB which has a PIC controller with a USB interface. A driver in RDOS provides a userland API to set references, read out temperatures, termomotor voltages, and to turn on and off relays. The heating pump is controlled through a Modbus interface using a USB to RS485 converter. The system also collects information from a solar inverter and a wind inverter and has a public web-page that shows history (http://home.rdos.net). The display on the portable PC is used to show current data in graphic form.

I also have the ADC system that can sample at 750 MHz and 600 MHz on two channels using an FPGA which then posts the data via PCI express to the main memory of a PC. I also have a fairly complicated userland software packet for signal analysis that can be used on this data. This software will take advantage of multiple cores by using multithreading and moving threads to their own core when possible. Among the functions is to measure power on distinct frequencies in variable intervals and phase difference between the channels which allows determining direction.

So as others suggested and seconded, I've splited up the page into two. Now we have three lists:

• OS Projects, a complete list with actualized "Status", I've left it as-is
• Notable Projects, list of mature hobby OS projects (developed mostly by one person), minimal "Status"
• Advanced Projects, list advanced and successful OS projects (developed by a team, has a user basis, sponsored by companies or universities etc. Also forks and clones belong here), minimal "Status"

In short, Notable Projects is from now on for hobby projects only, everything that's more than just a hobby should go to Advanced Projects page. I also think there should be no need to keep the status on these two pages up-to-date, a simple "Active"/"Discontinued" should suffice, as the OS Projects page has a very detailed status already (last commit, last release etc.)

Please review the modifications. I've added the OSes we've discussed. Yes, I've restored TempleOS too BTW, I've added Minix and ReactOS to the Advanced Projects page, because I felt they should be there.

(Just a sidenote, I really would like to see Managram and ToAruOS to be moved from "Notable Projects" to the "Advanced Projects" page )

Cheers,
bzt

Great! I think it's good to divide the OS projects into the three mentioned lists. And the lists look good on a first glance.

I also support your decisions on the projects you mentioned.

Greetings
Peter

Good idea to split into three.

I would say we should include AROS in the Advanced Projects, I used to be heavily involved with that one. https://aros.sourceforge.io

bzt wrote:About qword, I still vote for remove. The link goes to a repo that has nothing but a minimal kernel, there are no userland repos (except for an unfinished libc), not even a shell, and it's homepage is dead for more than half a year now (just reads "Server being setup come back later.") I honestly don't know why do you think it's self-hosting. A screenshot with gcc is not enough, because it has enormous amounts of build dependencies (python, pip, meson, nasm, autoconf, bison, etc. etc. etc). Don't get me wrong, qword looks interesting, and it should be listed under "OS Projects", it's just not ready to be listed in "Notable Projects" IMHO. Several OSes on the "OS Projects" list are more advanced and would deserve to be called "notable" more (in my humble opinion).

qword has a separate repo containing its build system. qword can run GCC, bash, coreutils and a few other things but it is not self-hosting as far as I know.

I do not know if the primary authors intend to develop it any further but I can contact them if that needs to be determined.

I've had a look at the pages, and they look good to me! The descriptions at the top seem nice and clear.

I'd like to see rdos on the advanced projects page, I think it qualifies. I recall @klange saying he'd like to keep ToaruOS as just a hobby: notable but not expected to be dependable, unless he wants to say otherwise. It's all up to the owners of each project, of course,* and I'm sure there will be some flexibility in which projects go on which page.

*: Barring complete nutcases, but hopefully we won't ever have one of those.