To me, this is more of a grab bag of random things rather than a well thought out argument for uniqueness. I still enjoyed it.
And it did make me think that I’d love to read an article that is about what made BeOS unique at the time (from one of us who was part of the community then) and also what makes Haiku unique now (from someone who is part of the community now). I’d love to read that article (or two articles if the authors worked together to make them a nice pairing).
The what made BeOS unique might be harder because you’d need to give a lot of background around why some things were revolutionary then for not just the tech but the effect that are fairly commonplace feeling now.
I was part of the BeOS community almost 2 decades ago, pretty much from the time they ported it to x86.
A few things that made it unique at the time:
64 bit filesystem that could be queried like a database;
incredible support for multiprocessing and low-latency that allowed for professional quality media editing apps;
an API that made programmers want to write software for it even when there weren’t that many users.
I keep a VM with the latest release of Haiku on it because someday, it will be mature enough for me to use as my daily driver. IT WILL. SHUT UP. IT TOTALLY WILL.
I love the versioned boot image idea. It’s the sort of thing you would hope never to need, but be really glad when it’s available.
This seems like a user experience bug:
When we create a contact in People, for example, everything we write in it are attributes. Notice the file size itself is ‘0 bytes’.
As far as I know, every OS that has a file system with forks/streams/xattrs does this. As the owner of a finite storage device, I’m looking at the file size to decide how much of my finite resource is used by that file. I don’t care whether your kernel technically puts the data in another fork, or the directory record, or some lookaside structure. I care that it puts the data somewhere.
As Linux gets more and more corporate and less targeted for the desktop, having a light-weight and responsive OS is enough to make it unique.
I do patch my Linux with the MuQSS scheduler, the best thing for Linux responsiveness, but I was recently told the Haiku one is essentially the same. This is awesome to me.
There is a lot and more in apps and hardware support that Heroku would need for me to switch over, but it seems like a cool project.
I do patch my Linux with the MuQSS scheduler, the best thing for Linux responsiveness, but I was recently told the Haiku one is essentially the same. This is awesome to me.
I don’t know a lot about the MuQSS scheduler, but from reading over the introductory document, it indeed looks pretty similar to Haiku’s. (I wonder where you read this previously, though?)
There is a lot and more in apps and hardware support that Haiku would need for me to switch over, but it seems like a cool project.
What would those be? Most minor tools are easily ported at this point.
IRC, oftc.net, can’t remember why I joined Con Kolivas’ channel #ck, but there. I consider him a friend after all this time and tested some of his prototypes way back.
Linux gets more and more corporate and less targeted for the desktop
Let’s hope the competitors get better in quality. I doubt I will want change to Haiku unless something really bad happens in the nix world, but hopefully its presence will make everyone else better nonetheless.
adding ever more complicated layers onto complicated layers to reinvent the wheel. most things should be done a few layers down, not by adding a few layers on top. this while having the same functionality 10 years ago, which most of the time was working as good as today, only less complicated and prone to break. the sound stack is just horrible, the most sane thing would be to throw out alsa and pulseaudio and use oss4, which implements most of the features. session and login management is also insane, a mess of daemons connected via dbus of all things. systemd people constantly reinventing square wheels (resolved, really?). while i’m at it, ps found a now one i didn’t know about: “rtkit-daemon”, fixing problems i don’t have, running by default.
I’ve been geeking out on schedulers for a long, long time and every encounter with vanilla Linux on a heavily-loaded box has been awful. It might behave better now, but that would be by very complicated code and bizarre special-case settings.
As a simple user, I just use the -ck patch set and ignore the horrors of the sound stack, systemd, Linux Foundation’s corporate politics, cgroups and what have you.
I mean, it kinda still works, but sometimes it feels the best desktop-experience parity with Windows was reached 20 years ago, if you exclude hardware support and games, and or with gnome3-type shit and everything got worse.
I’m not positive the desktop experience is as good as it gets but I am positive it’s no one’s priority.
I actually like Gnome 3 UI-wise, but the Linux scheduler seems to be more horrific than it used to be, and I remember it being bad a decade ago. I’ve had systems where X11 chugged hard and took 30 minutes to get to a vt when Firefox was stressing the system, when Windows on even more decrepit hardware was slow, but at least felt usable due to seemingly better scheduling - and it didn’t matter what WM you were using.
Yep, that application was released as part of the demo code package with the BeOS development kit. It still runs on Haiku indeed: https://github.com/HaikuArchives/3DMov
And especially when compared to various Gnu/Linux distributions that run X.org, or other systems that do likewise, there aren’t multiple layers all trying to mesh together. On the BeOS (and Haiku), everything is designed to work together in harmony
Does anybody know what specific problem this is referring to? And what “everything is designed to work together in harmony” actually means as a solution (Everything is consistent because it’s written by a very small group of people, perhaps?).
A system like a Mac is internally consistent; the layers were developed by the same people and mesh properly, and applications mesh together because they all follow the same APIs and HIGs and such.
Unless you install third party applications, in which case it varies between “follows the HIG to the letter, could be mistaken for an apple application” and “The title bar buttons are the normal Mac colours, but everything else is its own skin”.
I assume Haiku is trying to promote consistent guidelines, but the cynic in me wonders whether they would still achieve that if Haiku gained popularity and attracted lots of third party applications, either Haiku specific ones or ports from other platforms.
To me, this is more of a grab bag of random things rather than a well thought out argument for uniqueness. I still enjoyed it.
And it did make me think that I’d love to read an article that is about what made BeOS unique at the time (from one of us who was part of the community then) and also what makes Haiku unique now (from someone who is part of the community now). I’d love to read that article (or two articles if the authors worked together to make them a nice pairing).
The what made BeOS unique might be harder because you’d need to give a lot of background around why some things were revolutionary then for not just the tech but the effect that are fairly commonplace feeling now.
Anyway, enough babbling.
I was part of the BeOS community almost 2 decades ago, pretty much from the time they ported it to x86.
A few things that made it unique at the time: 64 bit filesystem that could be queried like a database; incredible support for multiprocessing and low-latency that allowed for professional quality media editing apps; an API that made programmers want to write software for it even when there weren’t that many users.
I keep a VM with the latest release of Haiku on it because someday, it will be mature enough for me to use as my daily driver. IT WILL. SHUT UP. IT TOTALLY WILL.
//It’s so cool.
It’s possible, but you should definitely write something up if you do.
I love the versioned boot image idea. It’s the sort of thing you would hope never to need, but be really glad when it’s available.
This seems like a user experience bug:
As far as I know, every OS that has a file system with forks/streams/xattrs does this. As the owner of a finite storage device, I’m looking at the file size to decide how much of my finite resource is used by that file. I don’t care whether your kernel technically puts the data in another fork, or the directory record, or some lookaside structure. I care that it puts the data somewhere.
Technically correct, the worst kind of correct.
FYI, versioned boot and rolling back to previous generations is also supported in NixOS.
Also similar facilities are available on FreeBSD, illumos, and SuSE.
As Linux gets more and more corporate and less targeted for the desktop, having a light-weight and responsive OS is enough to make it unique.
I do patch my Linux with the MuQSS scheduler, the best thing for Linux responsiveness, but I was recently told the Haiku one is essentially the same. This is awesome to me.
There is a lot and more in apps and hardware support that Heroku would need for me to switch over, but it seems like a cool project.
Does it do virtual desktops, btw?
I heard there was a better scheduler for desktop use. Didn’t know the name. Thanks for the tip.
It does.
The things that it’s missing that I would need to make it my daily driver:
Minimum:
Optimal:
I’d say BeOS is my favorite operating system of all time, but I can’t quite bring myself to say it since AmigaOS existed.
I don’t know a lot about the MuQSS scheduler, but from reading over the introductory document, it indeed looks pretty similar to Haiku’s. (I wonder where you read this previously, though?)
What would those be? Most minor tools are easily ported at this point.
IRC, oftc.net, can’t remember why I joined Con Kolivas’ channel #ck, but there. I consider him a friend after all this time and tested some of his prototypes way back.
The Godot engine would be one big thing.
Virtual desktops: yes.
Let’s hope the competitors get better in quality. I doubt I will want change to Haiku unless something really bad happens in the nix world, but hopefully its presence will make everyone else better nonetheless.
Disk encryption, does it have that? Password-protected screensaver?
BeOS had a password-protected screensaver.
How does mainstream GNU/Linux get worse?
NB: this turned out to be a poettering rant.
adding ever more complicated layers onto complicated layers to reinvent the wheel. most things should be done a few layers down, not by adding a few layers on top. this while having the same functionality 10 years ago, which most of the time was working as good as today, only less complicated and prone to break. the sound stack is just horrible, the most sane thing would be to throw out alsa and pulseaudio and use oss4, which implements most of the features. session and login management is also insane, a mess of daemons connected via dbus of all things. systemd people constantly reinventing square wheels (resolved, really?). while i’m at it,
psfound a now one i didn’t know about: “rtkit-daemon”, fixing problems i don’t have, running by default.i know, it’s open source, i can write a patch.
I’ve been geeking out on schedulers for a long, long time and every encounter with vanilla Linux on a heavily-loaded box has been awful. It might behave better now, but that would be by very complicated code and bizarre special-case settings.
As a simple user, I just use the
-ckpatch set and ignore the horrors of the sound stack, systemd, Linux Foundation’s corporate politics, cgroups and what have you.I mean, it kinda still works, but sometimes it feels the best desktop-experience parity with Windows was reached 20 years ago, if you exclude hardware support and games, and or with gnome3-type shit and everything got worse.
I’m not positive the desktop experience is as good as it gets but I am positive it’s no one’s priority.
I actually like Gnome 3 UI-wise, but the Linux scheduler seems to be more horrific than it used to be, and I remember it being bad a decade ago. I’ve had systems where X11 chugged hard and took 30 minutes to get to a vt when Firefox was stressing the system, when Windows on even more decrepit hardware was slow, but at least felt usable due to seemingly better scheduling - and it didn’t matter what WM you were using.
I’m not seeing Linux move away from the desktop at all. In fact I’m seeing more investment in the LInux desktop than ever.
It’s just that they’re investing in the wrong (from my selfish stance :) desktop environment :)
they’re moving away from the desktop and towards tablets, even though linux doesn’t run on any
Don’t have a machine where I can readily install and test now, but does haiku support this ol’ BeOS feature: https://youtu.be/s4A0Lc6DI7A?t=687 that I mimicked in this’ ol PoC (warning for lound noises) https://youtu.be/3O40cPUqLbU?t=161
Yep, that application was released as part of the demo code package with the BeOS development kit. It still runs on Haiku indeed: https://github.com/HaikuArchives/3DMov
[Comment removed by author]
Does anybody know what specific problem this is referring to? And what “everything is designed to work together in harmony” actually means as a solution (Everything is consistent because it’s written by a very small group of people, perhaps?).
A system like a Mac is internally consistent; the layers were developed by the same people and mesh properly, and applications mesh together because they all follow the same APIs and HIGs and such.
Unless you install third party applications, in which case it varies between “follows the HIG to the letter, could be mistaken for an apple application” and “The title bar buttons are the normal Mac colours, but everything else is its own skin”.
I assume Haiku is trying to promote consistent guidelines, but the cynic in me wonders whether they would still achieve that if Haiku gained popularity and attracted lots of third party applications, either Haiku specific ones or ports from other platforms.
GNU should be correctly capitalized.