Other XFS on BSD

perhaps it sheds some light on FreeBSD's utter community dysfunction.
It is a difficult balancing act for the project. You have two types of people (there are more I'm going to simplify).

  1. People like me who find many improvements unnecessary and either contribute to bloat, pointless redevelopment of existing workflows or take developer focus away from adding stability rather than new features.

  2. People like you who want fairly big changes to the operating system because you are truly not happy with how it performs in one area or another.

So far I see FreeBSD as doing a pretty good job of improving stability whilst keeping new features coming in. I don't personally dread new releases (unlike in Linux or especially Windows). Or to look at it negatively; "stability and correctness is slowly being eroded away whilst new features dribble in at a slow pace."

In short, if you really want new features daily, then you know Linux will be better for you going forward (super computers use Linux because features can be hacked and mashed in quickly to achieve benchmarks!). But I do recommend sticking around with FreeBSD because I simply don't see better anywhere else. OpenBSD is impressive in the stability department. But rarely sees new features. This is possibly more ideal for people like me but likely wont interest you.
 
The more security is important, the less number of packages you will have, because many of them won't fulfill your security requirements.
The more open source is important, the less number of package you will have, because some of them contain here or there a little piece of closed source software ie firmware blob or it might contain a non-free license.
 
kpedersen
Good stuff, very good. The bit you quoted, your two types of people.

"...utter community dysfunction". I strongly disagree with that part; these forums and the mailing lists don't indicate dysfunction to me. They show people trying to help others as best they can, sometimes strongly held opinions and beliefs lead to some rather entertaining threads (this thread has some ups and downs), mailing lists you get some awesome bikesheds going, but overall in my time of using FreeBSD (been around a while, since the 3.x era) any dysfunction is limited to a handful of people that just won't listen to others.

New features, drastic changes, easy if there is a huge paid staff, hard when there is a small volunteer "staff".
Anyone can add the features they want (either do it themselves or pay someone to) and then make the argument for inclusion in upstream.
Or if you want to, fork it. That's what Matt did for DragonFlyBSD. Strong feelings on both sides of the performance and scaling issue, so a fork and parting of ways.

ETA:
The above is my opinion, feel free to agree, disagree, call me names, don't call me names. Just remember how much you paid for my opinions
:)
 
Now that I'm not running as fast as I can to stay in one place anymore, I've noticed more and more the countless posts on Hacker News and other forums that bring up pointless topics about their new software or another countless rewrite of the same thing (that hardly anyone uses anyway) which drags in tens and hundreds of responses and I wonder what all these people are doing wasting their time with that crap.
 
  • Like
Reactions: mer
It is a difficult balancing act for the project. You have two types of people (there are more I'm going to simplify).

  1. People like me who find many improvements unnecessary and either contribute to bloat, pointless redevelopment of existing workflows or take developer focus away from adding stability rather than new features.

  2. People like you who want fairly big changes to the operating system because you are truly not happy with how it performs in one area or another.

So far I see FreeBSD as doing a pretty good job of improving stability whilst keeping new features coming in. I don't personally dread new releases (unlike in Linux or especially Windows). Or to look at it negatively; "stability and correctness is slowly being eroded away whilst new features dribble in at a slow pace."

In short, if you really want new features daily, then you know Linux will be better for you going forward (super computers use Linux because features can be hacked and mashed in quickly to achieve benchmarks!). But I do recommend sticking around with FreeBSD because I simply don't see better anywhere else. OpenBSD is impressive in the stability department. But rarely sees new features. This is possibly more ideal for people like me but likely wont interest you.

I think this is a gross oversimplification. I won't be wanting to retreat to GNU/Linux because I find many of the choices they make unacceptable, and as far as drastic changes goes, FreeBSD has gone through a lot.

The 13 release was the final straw. They gave up independence on a lot of things and it shows me the writing on the wall that FreeBSD is not for me anymore, and falls into the same category as GNU/Linux. Some of my stuff will move to NetBSD, others I will have no choice but to use GNU/Linux. it's frustrating because GNU/Linux has nothing I particularly care for -- the distros that are at least somewhat not brain damaged are rolling release, which I cannot do for my mission critical things. RHEL/SuSE/OpenSuSE/Ubuntu are particularly stable and have LTS versions, but they all have plague and taint beyond belief in base. I was an RHEL admin for years. I don't like bringing my work home with me.

I'm left in a bad spot on an island that is the 12.x series, and when that erodes away, I don't know what I'm going to do long term on everything other than NetBSD for what I can.
 
The attitudes of several users here and in the foundation thread is simply the wrong mindset... the fact some of them even praise OpenBSD, which is run by a narcissist (I like it when Theo takes apart idiots, sure, but his attitude is in the line of Torvalds -- utterly out of touch with things. I see some of myself in Torvalds and Theo, fwiw, so I'm self-aware) and has made dozens of questionable at best decisions underlines the way in which FOSS discussions often derail from facts, and into cult/religious musings.
Main parts of the showbiz are run by narcissists. And, does the audience care? Nope. What counts for them is the product, like a great movie, and not the assholish personality of an actor/actress, as long as he/she does not commit justfiable crimes or really bad things which are despised by society.

Same goes for software projects. Until I've got to work with people like Theo DeRaadt together (highly unlikely), I really give a shit about his personality or personality issues. What only counts for me is if the product, which by the way is not his baby alone, but there's a whole, dedicated team of developers behind it, fits the need I've got, or not.

And just like OpenBSD is not Theo DeRaadt's achievement alone, but a group achievement, the same goes for Linux and many other successful open source projects. So by narrowing down the project to just its most prominent public figure you are really trampling on all the other developers behind it, who are also doing tons of important and great work.

Since all these are group efforts, the argument like e.g., OpenBSD is BLAH because DeRaadt is a narcissist is oversimplification and just utter nonsense.

FreeBSD is an open source project. If you want a new feature, then it's either up to you to implement it or persuade/hire somebody who is able to do it. Furthermore if you would have done your homework you would have realized that the place to go for to do the second is not the forum here, but the FreeBSD mailing lists themselves. This here is just a small part of the community, and not representative for the rest of it at all.
 
This is a part of my make.conf, it speaks for it-selves,
Code:
OPTIONS_SET+=SNDIO
OPTIONS_UNSET+=PULSE
OPTIONS_UNSET+=PULSEAUDIO
 
Is this about the OpenZFS thing?

Yes and no. In general I think the move from illumos to ZoL and making ZoL THE reference implementation is crowdsourcing around a popular, but inferior, implementation. ZFS was always a "good enough for now" stopgap. I wanted hammer2 for years, but I learned that it would never be very portable and that voided my interest.

The topic here is XFS on BSD, so I disagree: a port of XFS is on topic.

The topic carries a specific connotation. FUSE is not that. In-kernel support is necessary. Fuck FUSE

Main parts of the showbiz are run by narcissists. And, does the audience care? Nope. What counts for them is the product, like a great movie, and not the assholish personality of an actor/actress, as long as he/she does not commit justfiable crimes or really bad things which are despised by society.

A narcissist can ruin something with their egotism.

Same goes for software projects. Until I've got to work with people like Theo DeRaadt together (highly unlikely), I really give a shit about his personality or personality issues. What only counts for me is if the product, which by the way is not his baby alone, but there's a whole, dedicated team of developers behind it, fits the need I've got, or not.

It matters when Theo's underlings are primarily Yes Men. They don't disagree with him. There's a hundred bad decisions OpenBSD has done, but among them:

Disabling SMT, removing LKM support, removing Linux emulation, refusing to adopt a journaling FS as even an OPTION, repeated removal of useful features/platforms because of narcissism etc.

FreeBSD is an open source project. If you want a new feature, then it's either up to you to implement it or persuade/hire somebody who is able to do it. Furthermore if you would have done your homework you would have realized that the place to go for to do the second is not the forum here, but the FreeBSD mailing lists themselves. This here is just a small part of the community, and not representative for the rest of it at all.

This is neither here nor there. I cannot change FreeBSD to be the way I want without forking it, and I think it's too far gone for that. Nobody would defect from FreeBSD to a smaller port, so it's meaningless, and we could only suck on FreeBSD's tit for so long before we'd need to go into our own direction.

And let me explain: Telling someone to "DO IT YOURSELF" is unhelpful. Not everyone has the time, specialization, skills etc. Ideas are indeed a dime a dozen, but being a jack of all trades means you lack specialization. Sure, I can work my way around C with compilers and libcs and kernels to some degree, but am I capable of doing everything an OS needs? No. Unfortunately.
 
And yet you still haven't provided objective reasons to adopt this filesystem, nor did you mention practical deficiencies given a set of use case scenarios. I can tell you've been spoiled by the GPL and like-minded individuals; your sense of entitlement is through the roof. That's once aspect of this community that I love, the entitlement is low; work ethic is high.
 
I accept the fact that there might be a few closed source binary blob firmware running on my system, eg my soundcard & videocard.
some of them contain here or there a little piece of closed source software ie firmware blob or it might contain a non-free license.

Firmware (open or closed) doesn't run on CPU, kernel address space, etc. It runs on the device's chip. Vendors should burn it into a ROM. They don't. So, the OS has to load it on a device. Firmware is not device driver (open or closed). Thus, there's nothing wrong with loading proprietary firmware (*). RMS/FSF propaganda machine would disagree with that, I suppose.

[EDIT](*)
-- you'll get it on device anyway, with or without help of an OS.
 
Firmware (open or closed) doesn't run on CPU, kernel address space, etc. It runs on the device's chip. Vendors should burn it into a ROM. They don't. So, the OS has to load it on a device. Firmware is not device driver (open or closed). Thus, there's nothing wrong with loading proprietary firmware (*). RMS/FSF propaganda machine would disagree with that, I suppose.

[EDIT](*)
-- you'll get it on device anyway, with or without help of an OS.
Sometimes we do get rather loose labeling for firmware vs. device drivers...
 
  • Thanks
Reactions: a6h
A narcissist can ruin something with their egotism.

It matters when Theo's underlings are primarily Yes Men. They don't disagree with him. There's a hundred bad decisions OpenBSD has done, but among them:

Disabling SMT, removing LKM support, removing Linux emulation, refusing to adopt a journaling FS as even an OPTION, repeated removal of useful features/platforms because of narcissism etc.

AFAIK OpenBSD is far from dying, and doing quite well. OpenBSD's main thing is security without compromises, and that means they are taking more drastic actions compared to other OSes in order to achieve it. You can like it, or not, either you'll use it or leave. Simple as that. You instead would like to see them do things in a different way, because you dislike the direction the project is moving to. This is why you call DeRaadt a narcissist, because how dare they not doing the things your way. And this is quite being full of yourself from your side.

Reg. disabling SMT, even well known experienced AMD64 architecture kernel developers from the Linux camp agreed that it was the right thing to do, because the whole thing is a clusterfuck of a hardware bug which can only be solved by new CPU designs maybe:

View: https://www.youtube.com/watch?v=jI3YE3Jlgw8


Removing kernel modules also makes lot sense from a security perspective, because this is really quite a big attack vector and if something cannot be loaded during runtime it cannot be abused to log stuff on your system or do bad things. This makes the potential for kernel based root kits much much smaller.

And so on... journaling, well they've got the same stance like FreeBSD. And aside that OpenBSD was never meant to be the best performing OS ever, in fact in many areas in the past it was quite bad, but the most secure out of the box instead and during normal operation.

And let me explain: Telling someone to "DO IT YOURSELF" is unhelpful. Not everyone has the time, specialization, skills etc. Ideas are indeed a dime a dozen, but being a jack of all trades means you lack specialization. Sure, I can work my way around C with compilers and libcs and kernels to some degree, but am I capable of doing everything an OS needs? No. Unfortunately.
It doesn't matter if it is unhelpful, what I just explained is the basic principle of open source: if you want a feature, either implement it yourself or persuade/hire somebody to do it for you.

Let's just say the way you are trying to persuade people is not exactly the best way to handle it, because you're doing it in the wrong places.

Probably the best thing for you is if you really want XFS for a RDBMS partition to move over to Devuan, there you can have it instantly.
 
Let's just say the way you are trying to persuade people is not exactly the best way to handle it, because you're doing it in the wrong places.
Sometimes we do get rather loose labeling for firmware vs. device drivers...
And yet you still haven't provided objective reasons to adopt this filesystem, nor did you mention practical deficiencies given a set of use case scenarios.
Not going to disagree with any of these :)
 
And just like OpenBSD is not Theo DeRaadt's achievement alone, but a group achievement, ...

Since all these are group efforts, the argument like e.g., OpenBSD is BLAH because DeRaadt is a narcissist is oversimplification and just utter nonsense.
I actually consider OpenBSD to be a great product, and very good at what it does. It has a relentless focus on high quality, achieved through clean and careful implementation, and removing unnecessary features. This leads to excellent security, better than anything else in the open source and full feature space. Why does this work? Because there is a BDFL (benevolent dictator for life) who can set the project's culture. Now, Theo's way of interacting with people is not always benevolent, he can be quite harsh. But one of the significant differences between him and for example Linus is that he remains reasonable when being harsh, he doesn't have to go to anger management classes after having strong opinions, and the sponsors of OpenBSD don't but him in a penalty box.

By the way, this does not mean that OpenBSD is the operating system that a random user needs. It is specialized, it lacks lots of features (some of them are features I need, which is why I currently don't run it on my main server). But it is very good at what it does. One of the wonderful things about free software is that it frees the developers from having to try to serve all their customers badly (which is what typically happens in commercial software development), they can choose to serve one particular set of users well.

ZFS was always a "good enough for now" stopgap.
That statement might be true for you. For much of the rest of the storage industry, ZFS is the best solution in the open/free space. In that space, it is the only file system that has two features that I consider vital: integrated RAID and volume manager, and full checksums. You can get that in commercial offerings, but not otherwise in free software.

Sure, I can work my way around C with compilers and libcs and kernels to some degree, but am I capable of doing everything an OS needs? No. Unfortunately.
You are free to give the FreeBSD foundation the money required to implement XFS. My educated guess would be that this would be expensive, and so perhaps you don't want to do that. You are also free to convince people that implementing XFS on FreeBSD would be a good idea, in the hope that you can inspire volunteers and donors. You have been doing a spectacularly bad job at that on this forum.

The way FOSS software works is this: Something gets done if either money or manpower show up. Today, for major projects (in particular in Linux) that is usually a commercial sponsor, a big company (IBM, Oracle, NetFlix, Google, ...) that needs some feature, and pays for it to be written. Or big companies sponsor the various foundations, which then employ developers (for example Greg Kroah-Hartman for Linux), or they internally have open source developers on staff (IBM used to have literally thousands of them). Minor projects are done by volunteers (often college students, for example over the summer). Long-term efforts are sometimes done by software engineers employed in the industry, in their spare time, often with partial support from their employer (for example vim maintenance and leadership).

Implementing XFS on FreeBSD is not something a few volunteers can do quickly; it would be a multi person-year project. It would also benefit from a professional staff; college students typically don't have the experience required to build kernel-internal file systems. So funding would need to be found.

EDIT: Changed "Porting XFS..." to "Implementing XFS...", to be clear with what hardworkingnewbie wrote below: Getting native XFS support is not a port, since the existing source code can not be used for the purpose.
 
Porting XFS to FreeBSD would not just be a port, since the Linux kernel and therefore XFS in it is under the GPLv2, which we all do know is incompatible to the BSD license of FreeBSD.

So this would be not just a port, but in reality a rewrite from scratch, which is way more labour intensive than just porting it over. This is definitely not something an unexperienced developer can simply pull off. A suitable developer would need to be proficient enough in these areas:

* file system development
* Linux kernel development and
* FreeBSD kernel development.

So there are not so many developers around for sure which might do that.
 
A suitable developer would need to be proficient enough in these areas: ...

And to make sure that proficient developer does not inadvertently copy existing GPL-ed source code, they would pretty much never have worked on the Linux kernel file systems, in particular not that XFS implementation. Otherwise, the owners of the Linux copyright (hmm, who might that be?) could and probably would sue, claiming that the new implementation is a derivative work, and therefore under GPL.

So now try to find a group of people who are experienced kernel developers with file system experience, but have not worked on Linux. Good luck!

P.S. I edited my post above to clarify that XFS can not be ported into FreeBSD, only re-implemented.
 
And yet you still haven't provided objective reasons to adopt this filesystem, nor did you mention practical deficiencies given a set of use case scenarios. I can tell you've been spoiled by the GPL and like-minded individuals; your sense of entitlement is through the roof. That's once aspect of this community that I love, the entitlement is low; work ethic is high.

Because statistics don't exist that satisfy you or anyone else's burden of proof. I can provide phoronix articles putting it up against Linux FS and you say that's biased or not good enough.

I think UFS is a nightmare to maintain and lacks the performance necessary to compete, and the attitude you and others have said hasn't convinced me otherwise. It reeks of pure fucking propaganda/hivemind mentality.

AFAIK OpenBSD is far from dying, and doing quite well. OpenBSD's main thing is security without compromises, and that means they are taking more drastic actions compared to other OSes in order to achieve it. You can like it, or not, either you'll use it or leave. Simple as that. You instead would like to see them do things in a different way, because you dislike the direction the project is moving to. This is why you call DeRaadt a narcissist, because how dare they not doing the things your way. And this is quite being full of yourself from your side.

Red Herring. I didn't say it was dying. I said it made questionable decisions. I called deRaadt a narcissist because his attitude would be unacceptable in most FOSS projects. Regularly abusive and foul language directed directly at others. There's a difference between telling someone "This is a stupid argument" vs "You're a stupid idiot" -- you can google and see theo's regular verbal abuse of people in mailing lists.

As for why I don't participate in mailing lists, I have my personal reasons. I have a history with FreeBSD, stretching back years, so don't assume I have the same privileges and capabilities you do.

Probably the best thing for you is if you really want XFS for a RDBMS partition to move over to Devuan, there you can have it instantly.

RDBMS is not the main reason I want XFS. I want XFS to replace the trash that is UFS because UFS is an unreadable, unmaintainable nightmare from the 1970s. You're free to disagree, but nobody is using FreeBSD in a professional capacity with UFS -- most use ZFS. I can't and won't use ZFS anymore.

I actually consider OpenBSD to be a great product, and very good at what it does. It has a relentless focus on high quality, achieved through clean and careful implementation, and removing unnecessary features. This leads to excellent security, better than anything else in the open source and full feature space. Why does this work? Because there is a BDFL (benevolent dictator for life) who can set the project's culture. Now, Theo's way of interacting with people is not always benevolent, he can be quite harsh. But one of the significant differences between him and for example Linus is that he remains reasonable when being harsh, he doesn't have to go to anger management classes after having strong opinions, and the sponsors of OpenBSD don't but him in a penalty box.

I don't think anyone should be forced from a project for their attitudes. I simply strongly disagree with OpenBSD -- that doesn't mean they're not important for OpenSSH, Game of Trees, or other advancements for the FOSS community.

When people were trying to cancel Stallman, including the HardenedBSD dev Lattera, I was facepalming. He said some pretty bad comments, but while I dislike RMS strongly that's his project and his legacy, and it's not my place to force him from something he started.

That statement might be true for you. For much of the rest of the storage industry, ZFS is the best solution in the open/free space. In that space, it is the only file system that has two features that I consider vital: integrated RAID and volume manager, and full checksums. You can get that in commercial offerings, but not otherwise in free software.

I think ZFS has valuable features, and compared to Btrfs it's a much better filesystem for reasons I've stated. I cannot however condone its recent development track.

You are free to give the FreeBSD foundation the money required to port XFS. My educated guess would be that this would be expensive, and so perhaps you don't want to do that. You are also free to convince people that porting XFS to FreeBSD would be a good idea, in the hope that you can inspire volunteers and donors. You have been doing a spectacularly bad job at that on this forum.

I knew by the time drhowarddrfine attacked me unfairly on the foundation forum post that it wasn't gonna happen.

I mostly stay interacting here on this particular topic because people are being ignorant. I'm not interested in leading the project, nor do I have the ability to. I said that a few times before, for very reasonable concerns. Nor would I want to secure donations. The detractors like Beastie7 discard everything I say because it won't meet an impossible burden of proof. This is cult think. It's the same thing you see in other statements.

Porting XFS to FreeBSD is not something a few volunteers can do quickly; it would be a multi person-year project. It would also benefit from a professional staff; college students typically don't have the experience required to build kernel-internal file systems. So funding would need to be found.

Volunteers can do it based on the docs. I would have joined an effort, and still probably would, as a gap-filler. I'm not interested in getting paid, nor do I believe that more than a year of effort would be required by a small team to get a working driver. XFS is not some black box like ext or reiser or whatever.

Porting XFS to FreeBSD would not just be a port, since the Linux kernel and therefore XFS in it is under the GPLv2, which we all do know is incompatible to the BSD license of FreeBSD.

It's a reimplementation.

So this would be not just a port, but in reality a rewrite from scratch, which is way more labour intensive than just porting it over. This is definitely not something an unexperienced developer can simply pull off. A suitable developer would need to be proficient enough in these areas:

* file system development
* Linux kernel development and
* FreeBSD kernel development.

So there are not so many developers around for sure which might do that.

You clearly didn't read the documentation I linked in the thread this originated from. It describes the structure of an IRIX-era XFS (which other than directory structure versions, is similar to the modern Linux XFS) You don't need to know Linux kernel driver development, and as far as filesystems go it's far far simpler than either the FreeBSD UFS version, or EXT3/4 which are the standard for many linux distros.

I'm done here. That's all I can say. Spout all the bull and lies you want. Clearly, this exposes the rot that FreeBSD's community is suffering. Never in a million years did I think the Linux community toxicity would spread into anything but OpenBSD, but here we go.
 
so don't assume I have the same privileges and capabilities you do.
The whole point about open-source and free software is that we (users) all have the same privileges and capabilities to access the software. So you do have the same privileges as the rest of us and always will so long as the project is under free licenses. Note: As a developer however, it is often a meritocracy so only doing actual work will improve your standing in that part of the community

Never in a million years did I think the Linux community toxicity would spread into anything but OpenBSD, but here we go.
Seems a strangely pre-emptive thing to say when OpenBSD is renowned for being a community that is the least likely to engage with beginners pushing for ideals that go against their own. They will simply explain why the newcomer is wrong, and then ignore them.
 
I'm done here. That's all I can say. Spout all the bull and lies you want. Clearly, this exposes the rot that FreeBSD's community is suffering. Never in a million years did I think the Linux community toxicity would spread into anything but OpenBSD, but here we go.
Dude, you've got serious personality issues when you confuse a normal discussion about technical stuff with what you're saying here. Seriously, get some help.
 
Back
Top