I'm highly fascinated between Haiku and ReactOS, but I havent had a chance to use both in a while. I want them to become available. I really appreciate that Haiku was meant to be a remake of BeOS (iirc) and it does a fantastic job. I feel like we don't do enough OS research like we used to. We've started to dive into making new languages like Rust and Go, but I guess the next wave I'm waiting for is new and interesting Operating Systems.
I feel projects like MirageOS[0], Qubes OS[1], and unikernels in general are laying a lot of groundwork for interesting new developments in operating systems. And as a bonus, there's a lot more automated work on security and safety!
[0] https://mirage.io (and cool sub projects like running Mirage on a ESP32 with reprocessing to make a homebrew gameboy-like platform [0.1])
I’m stoked that unikernel papers are still popular (and there’s still a lot of unexplored territory), but it’s not exactly a groundbreaking area!
I share the original post’s sentiment about the lack of exciting operating systems research around topics outside of the standard areas like file systems, memory management, and security. Especially since these issues, for many people, are secondary to the perceived features of operating systems (like drivers and widget toolkits). I’d love, for example, to see some truly fresh ideas for new or different system calls (in addition to Fuchsia and Haiku).
If you're interested in contributing I'd look into it a little closer as I don't think containers align with the security model of Qubes. They'd rather support spinning up single container VMs which run separate Docker instances... probably not even in a Docker swarm due to the security channels that opens.
We're definitely not doing enough interface research anymore, now that our corporate overlords have decided that "apps" are the thing to aim for, as they're easier to exploit for maximum profitability. At one time, more integrated approaches were considered. It's sad that Unix pipes and all that's wrong with them are one of the few remainders on this particular field.
I still have a BeOS 3.5in floppy in a drawer somewhere, so this announcement produced a little nostalgia for ~1999 or so. That said, it's sort of hard to imagine lots of new OSes springing up, they take a lot of care and feeding... I would think the obvious choice would be to somehow to support packages from another OS (say Snap or Flatpak apps from Linux), so that you could overcome the need to port everything (or have no users).
That does sound a bit worrisome if it indeed were to become an OS that runs on every platform including desktops. If they ever close source it, then I will be really worried.
IIRC, it is intended to replace both Android and Chrome OS. And as shown by the Android Open Source Project, they don't need to close-source it to engage in worrisome behavior.
Useful OS will probably never append , when you see how long it took to create Linux, thousands of people that worked on it for a very long time and still it's not on part with Windows for desktop.
To be fair, most of this work, as is the case for any popular operating system, is mostly drivers. It’s entirely possible to write a succsssful operating system for a narrowly defined platform.
Well, Haiku reuses FreeBSD's ethernet/WiFi drivers, and we will probably wind up reusing DragonFlyBSD's DRM drivers. Most of our time in kernel development is actually spent on the kernel, not in driver development.
Except I wouldn't call Haiku or ReactOS "research projects." Indeed, one of their goals, whether explicit or implicit, is to provide an alternative to Windows on the desktop.
There are a few of these. They're an "alternative to Windows" in the same sense that riding a penny farthing is an "alternative to travelling by bicycle" or sending a telegram is an "alternative to using the telephone". It's just nostalgia.
Whether these systems could attract the same development effort (and money, though what if anything Haiku Inc. does with the money is pretty unclear since they're lousy at basic organisational responsibilities of a charity) if they admitted they aren't a serious "alternative to Windows on the desktop" is an open question.
> Haiku, Inc. publishes yearly financial reports detailing financial transactions
One draft report for 2017, misfiled in the hopelessly mismanaged Haiku Inc. "website" after many years since the last one (in 2010) does not count as "yearly financial reports" and since this is a one page summary it can't be said to be "detailing financial transactions" either.
Mostly decisions aren't made at all, anywhere. That financial report you're calling "yearly" took months for the Haiku Inc. board not to agree. That's right, months, not for an agreement, but for the _lack_ of any response whatsoever to be taken as tacit permission to just publish it anyway.
Ah yes, a board made up of 5 people who work on this in their spare time is expected to be as organized as multimillion-dollar nonprofits. /s
Sarcasm aside, I don't know what decisions you're referring to which "aren't made at all, anywhere" -- the board agreed to a cost increase for server hosting earlier this year so we could switch providers.
You put "website" in scare quotes, and I don't know why -- in fact they deployed a new website last week, which is much better organized than the previous one was.
The point is, Haiku, Inc. is not as responsive as we'd like, indeed. But the money isn't going into a black hole, and they do pull together to get things done when we really need it.
That's the fifth time I've downloaded a USB live disk image, written it to a USB key, booted, and had it panic out with "no bootable partition" in the first four seconds.
Each of those dozen or so has different causes. So until you file a report with a syslog attached, it's impossible to say whether or not yours is a new one or a duplicate.
But if it is a duplicate, and you already know about the existing reports, why are you frustrated that you've tried again and nothing has happened? We are pretty diligent about managing the bug tracker and closing tickets when they're fixed, so if the ticket is still open, it's probably not fixed.
Likely if you are booting from USB, the problem is your USB3 controller. Try booting from a USB2 port, or booting from another medium.
Display flickering/lag/etc. sounds like the intel_extreme driver being broken; there is a ticket about this also. You can avoid that for the time being by blacklisting the driver or forcing "fail-safe graphics" (VESA).
Try a different key. I've had a couple totally ok and error free usb keys literally refusing to boot when installing Linux on a PC. Using a different usb drive solved the problem, and the dongles which refused to boot on that pc worked perfectly on different ones.
I've also seen laptops which wouldn't connect to their home router WiFi no matter the settings, I had to put either an external WiFi adapter or swap the router, needless to say that both the WiFi cards and routers were 100% functional when connecting to other hardware.
Some pieces of hardware just don't like talking each other.
I used BeOS for a little while, so I keep track of Haiku every quarter or so. It’s missing two things for me to use it again, assuming the wireless drivers are done now: Power management (including suspend/resume) so I can set it up on an old netbook (for which it would likely be perfect) and an ARM port for the Raspberry Pi (which are the only “desktop” machines I have available).
I like Haiku. In my experience, it has most responsive window manager for any operating system that runs on 90s-00s hardware. I think it has a lot of potential, but presently, I’d describe it as an alternative to something like Chrome OS. I wish the web browser situation was comparable to Linux, macOS, and Windows.
I should admit that I have an academic interest in operating systems and Haiku, for a variety of reasons, is exciting (e.g. it’s a functional open source operating system that isn’t strictly POSIX).
Except... Haiku is POSIX? The POSIX compatibility isn't just some compatibility layer, it goes all the way into the kernel. We use the fork-based process model, POSIX filemodes, etc.
What does "strictly POSIX" mean, and how does Haiku not meet it?
If there are POSIX APIs we don't implement, it's simply because we haven't gotten around to implementing it / nothing needed it so we didn't bother, not because we are explicitly avoiding said APIs.
packagefs aggressively caches decompressed data and builds a bunch of indexes in RAM, and that's where the larger requirement comes from. It's not very well optimized for memory, though, probably someone dedicated enough could bring the requirement back down to 128MB.
While optimization is indeed the hallmark of Haiku, the above difference does not matter much these days, of course. What could be a big deal, though, is the (missing) support for popular platforms with somewhat limited resources, such as the Raspberry Pi. Would not it be nice to have a smart TV or a monitor with Haiku built in?
It would indeed. Various contributors have been working on the ARM port for almost a decade now, and it's just stalled due to a lack of developer time for the most part.
IIRC, the current status of the ARM port is that the kernel is "mostly done", and at least one of the SDHCI or USB bus driver need to be ported (or written, in the case of the rpi) so that the boot can continue past the kernel.
I run Haiku in a VM. Can I upgrade it from within Haiku or do I have to nuke what I have and start over? I haven't done a lot of customization but enough to make it a pain to do over.
If you are on the beta release or a recent nightly image, you can just run SoftwareUpdater and it will upgrade to the most recent build on either branch.
If you are on an older (i.e. pre-2017) nightly, you can still upgrade automatically, but you'll have to manually change package repositories to do so. If you are on an older release (alpha4, etc.) then you can't upgrade at all, you'll have to reinstall.
Going forward, it will be possible to upgrade from beta1 directly into beta2, or whatever the next release is.