Only Microsoft can give open source the gift of NTFS. Only Microsoft needs to

We concentrate on their technical aspects, but file systems can get pretty political. They're one of the last fronts still fighting in the Interoperability Wars. While you can plumb any number of open file systems to Linux if you need what they have, NTFS remains a problem.   Why? Because it's a very practical issue that can't …

[TonyJ]

Even if they did...

...you can bet the first comments here would be "I haven't used anything Micro$oft for decades and wouldn't start now" types etc etc

But, joking aside, the article is right - these days NTFS isn't (shouldn't be) a battle ground. It's winning them no new business - no one, ever, is going to choose Windows because of NTFS*. It's just another filing system at the end of the day.

Well ok, I guess you are if, like the article states, you need to plumb into an existing NTFS infrastructure but that's a shitty way to have to justify an operating system choice.

[An_Old_Dog]

Using NTFS

I use NTFS on flash drives and portable hard discs because:

(a) I've found it to extremely-durable against PC lock-ups and dirty shutdowns, and

(b) It's cross-OS-portable, readable by Windows, Linux, Unix, and OS X. The computer-world I deal with contains all four of those OSes.

[bombastic bob]

Re: Using NTFS

how does EXFAT compare to that? Just curious...

[Richard 12]

Re: Using NTFS

ExFAT doesn't have a journal, so you're more likely to lose data.

Though IIRC it uses alternating FATs so you're likely to retain everything you didn't touch. Unlike FAT32, with the single table that dies and the entire stick is gone.

[rajivdx]

Re: Using NTFS

ExFAT has a journaling version called TexFAT, but journaling works only on supported OS's like Windows CE, on all other OS's it looks like plain ExFAT.

All FAT versions including FAT32, FAT16 and FAT12 had 2 FAT's. If the drive was pulled out before the transaction was completed then the FAT's did not match and the OS could attempt to 'roll back' the drive to the last known good state.

[alcalde]

Re: Using NTFS

I have the same needs, and I use UDF. NTFS, as noted, isn't particularly portable at all. It works on Linux if you install the FUSE driver, but the journaling format was never reverse engineered so if the file system is in an unclean state you're stuck. According to Wikipedia, "Native NTFS write support is included in [OS X] 10.6 and later, but is not activated by default, although workarounds do exist to enable the functionality. However, user reports indicate the functionality is unstable and tends to cause kernel panics." Nice.

UDF is more portable and less problematic (NTFS' propensity for fragmentation is just ridiculous).

[bombastic bob]

Re: Even if they did...

Well the best use for the NTFS driver in Linux, In My Bombastic Opinion, is for a bootable Linux image that assists with data recovery and backups.

Open source disk imaging and file-by-file recovery utilities on a pre-built Linux bootable CD/DVD image. It has been done (of course) but an up to date kernel NTFS driver makes it all that more reliable and valuable.

I would like very much to be able to mount a "ghost image" of a hard drive (stored as a disk file, let's say), and then pick individual files off of it and recover them, or to do a file-by-file comparison between current and last ghosted image (particularly useful for something like a ransomware recovery process). LOTS of potential uses here in data recovery. Or, generate the ghost images to begin with, via the same "recovery tools" CD/DVD.

[Enric Martinez]

Re: Even if they did...

I use NTFS partitions for data I want to have available for both OSses in dual-boot PCs.

I would be pretty happy using NTFS as a Linux root partition, this would make it possible, for instance, to have one root with both kernels and put drive space to use that would otherwise just sit there inactive when you have booted into either OS.

[ElRegioLPL]

Re: Even if they did...

You arent joking though

They are all lunatics, you wonder how they survive in the real world. Walk into a job interview telling the interviewer their hate for Microsoft or "Micro Dollar Sign oft, ha ha!"

[D@v3]

Re: Even if they did...

no joke, there is a guy where i work, (he isn't in IT), and CONSTANTLY refers to them as MickySoft. Thinks he's really clever.

[bazza]

Re: Even if they did...

Battleground or not, it would cost MS money to resource an OSS and maintained version of NTFS, and there is simply not the business there to warrant it.

A real question is how much maintenance does the formerly Paragon code actually need?

[Plest]

Re: Even if they did...

I used to be anti-MS, they got on my nerves but I made an effort to find bit of MS tech I liked and I was forced to learn more and more of it and to be honest it's all just tech now. Maybe I'm just a sad old fart at 50 but the older I get the more I pick and choose battles worth fighting and MS over Linux over OSX over Android over Plan9 over...it's just not worth it anymore.

Live and let live, if you like it and find it useful then use it, else just make a not of why you don't like it, fix it if you can, else move on 'cos life's simply too short for some things to stress you out.

[VoiceOfTruth]

It's the same old story with Linux - it's just one more thing

It's always been this way. It was Word format documents. It was SMB/CIFS. It's NTFS (and has been forever). It will be something else tomorrow.

Anyone seeing the response to the release of the Paragon code should take note. It was a thoroughly unwelcome attitude. Miserable ungrateful sods who can't do the job themselves but who think the world owes Linux something chimed in, and they repeated some of that here in a previous article. Some people even complained that because the code was once (and a version still is) closed source, it was bad. As I wrote - miserable ungrateful sods.

-> The company was founded in Russia and he's Russian, neither of which helps after Vladimir Putin's invasion of Ukraine on February 24

Yeah. Where were you when the USA illegally invaded Iraq and murdered 1,000,000 people? Why not bring the same attitude to source code from the world's number one prison camp?

-> Microsoft. Here's your chance. Do a good thing.

Linux, here's your chance. Do a good thing. Get rid of the GPL and use a friendly licence rather than that viral piece of crap.

[Cederic]

Re: It's the same old story with Linux - it's just one more thing

It's fine, you're welcome to use something other than Linux. Clearly you're not a fan of its open source model or its licence, both of which have helped it become the world's most successful operating system.

Incidentally your name is ironic when you claim a million people murdered in Iraq.

[VoiceOfTruth]

Re: It's the same old story with Linux - it's just one more thing

Linux is not an operating system, it is just a kernel. And so say all the pedants. It's funny that Linux servers depend on non-GNU things to operate. OpenSSH or OpenSSL? Hmmm. When it suits Linux people they use plenty of non-GNU software.

How do you measure success? If it is in terms of income and profit, both Windows and macOS beat Linux by humungous margins. If you measure it by giving up your time, effectively valuing your time at zero, then Linux is a big winner for companies like RedHat. A fool and his time are soon parted. Thanks for all them commits.

[bombastic bob]

Re: It's the same old story with Linux - it's just one more thing

I do not think you should lump all Linux users into the GPL-fascism camp...

Linus (apparently) chose GPLv2 for the kernel and the tool makers pick whatever licenses they want. Distros put binary packages together for download (etc.) and make the source available to comply with the licenses.

And only a few of these distro makers even TRY to be "GPL purists". Debian-based distros have boatloads of "non-free" packages available. Works for me.

(I guess calling it 'non-free' is enough of a virtue signal for the purists, but it ALSO does not stop people from installing them easily)

[VoiceOfTruth]

Re: It's the same old story with Linux - it's just one more thing

Your comment is fair. It does seem sometimes like the purists shout the loudest. Sometimes that does reflect on the Linux thing as a whole. The absolute spitefulness of some people is hard to avoid, and I won't ignore it.

[georgezilla]

Re: It's the same old story with Linux - it's just one more thing

... and I won't ignore it. ... "

Why not ?

You seem to be doing a pretty good job of ignoring reality.

" ... Sometimes that does reflect on the Linux thing as a whole. ... "

Only to ignorant people.

[Anonymous Coward]

Re: It's the same old story with Linux - it's just one more thing

"The absolute spitefulness of some people is hard to avoid"

Oh Kettle thou'rt black, quoth the Pot.

[martinusher]

Re: It's the same old story with Linux - it's just one more thing

People who dump on GPL tend to not be old enough to remember why GPL exists in the first place. Going back to the early days, the Wild West days, it was common practice for corporations to take open source code and put their own copyright banners on it, claiming ownership. This was a nuisance. GPL stopped this practice dead in its tracks.

Is it perfect? No. Could it be better? Yes. But understand the motivation before dumping on it -- and if you think you can do better, go for it. That's the essence of open source.

[VoiceOfTruth]

Re: It's the same old story with Linux - it's just one more thing

The problem with so many GPL supporters is they think other people don't know about GPL and why it exists and they feel a need to explain it.

[Alan Brown]

Re: It's the same old story with Linux - it's just one more thing

There is STILL a huge body of folk out there who think that "open source == public domain"

This is a particular issue with Chinese students I deal with, to the point that I have a couple of printed handouts explaining the legal positions,and that anyone caught stripping GPL (or other) attributions will earn an automatic removal from the organisation without further warning

[eionmac]

Re: It's the same old story with Linux - it's just one more thing

I would much appreciate a link to your 'printed handouts; (URL or PD or whatever); as find some folk here in UK have the same attitude as you impute to Chinese students

[Peter Gathercole]

Re: It's the same old story with Linux - it's just one more thing

There's nothing wrong with a mix-and-match model of software provisioning, just so long as you keep within the licenses.

Just because OpenSSH and OpenSSL are non-GPL licensed does not mean you can't use them on a GPL platform. Hell, the Open Software Foundation even made provision for that by making a large part of the GNU/Linux tool-chain available under the LGPL, which is less restrictive about building tools to run on GNU/Linux.

There's also nothing to prevent a software supplier to create non-GPL proprietary paid-for software to run on GNU/Linux. It's just that some bigots spread FUD that suggests that anything running on GNU/Linux has to be published under GPL. It's not, and never had been, the case.

What GPL was designed to do was to prevent someone (I'm not naming any names, but certain large companies have been proved to do this) taking all the work put in to create software from being incorporated into their product, and then possibly IP protected via patent or copyright. It's this appropriation that GPL is to prevent.

It has a secondary purpose to try to encourage other people to produce Open Source products, and I believe that it is these clauses that you're most against. Some of these clauses, particularly in GPLv3, can make it a bit daunting to write software, until you really understand them.

I think you really ought to look at Red Hat's business model. Yes, they are benefiting from curating a GNU/Linux distribution, but the only real financial benefit is the support and training they offer on the GPL components of RHEL, and the licenses for the proprietary tools that they ship.

But looking at their proprietary additions, there's no difference in doing that than if you were to produce management tools that can be used on top of, say, Windows.

And Red Hat are one of the largest contributors to the Linux kernel, and to some of the other GPL tools in the repositories. So it's not a one-way street.

[heyrick]

Re: It's the same old story with Linux - it's just one more thing

"It's this appropriation that GPL is to prevent."

It swaps one appropriation for another in its total inability to play nice with other recognised open source licences. It would be one thing if GPL code was required to respect the rules of GPL, but requiring everything in proximity to do so as well is overreacting.

Additionally, it is a horribly written licence. The actual terms of the above are unclear when applied to that which isn't Linux (and even within Linux has been subject to arguments about how and when it applies). The GNU FAQ basically absolves them of responsibility by saying that the courts will decide. Screw that, I'm not a lawyer, I don't want something that may be decided at some future date in a way different to my interpretation . . . or worse, decided differently in different states/countries none of which are my own.

It's just a bad licence wrapped up in far too much shouty politics. But the underlying concept is good and open source has made the impossible happen.

[VoiceOfTruth]

Re: It's the same old story with Linux - it's just one more thing

-> It swaps one appropriation for another in its total inability to play nice with other recognised open source licences.

That is so.

[georgezilla]

Re: It's the same old story with Linux - it's just one more thing

" ... Additionally, it is a horribly written license ... "

Then don't use it.

It's not a hard concept to understand.

Well apparently to some it is.

The only time that you have to release something under it is if you use something that already is.

Don't want to release your code under it?

Then don't.

Just don't use any code that is, in yours.

Again ...

It's not hard to ...........

Well shit!

< heavy sigh >

FFS!

< shakes ,head >

[SImon Hobson]

Re: It's the same old story with Linux - it's just one more thing

It swaps one appropriation for another in its total inability to play nice with other recognised open source licences.

Err ? It plays quite nicely - taking a commonly cited example, OpenSSH isn't GPL but Linux and OpenSSH play very nicely together thank you very much.

Now it is true that there are incompatibilities - and this comes up with (for example) ZFS. Is it the GPL's fault that ZFS (Sun's implementation) is released under a licence that's not compatible with the GPL ?

Additionally, it is a horribly written licence. The actual terms of the above are unclear when applied to that which isn't Linux

Rubbish, it's actually really easy to understand - at least at version 2 which I think is still the prevalent version. Really really simple :

If you give someone a binary built from GPL code then you also have to give them (if they ask for it) the source. If you take some GPL code and modify or expand it, then distribute that code, then you can only do so under GPL (you can't take code under GPL, and re-purpose it under a different licence). You can't add limitations/extras to the licence - so you can't remove those basic freedoms when you pass on the code.

That's basically it - what's hard to understand ? Of course, if your desire is to take someone's code, create a derivative work, and then sell that under a restrictive licence - tough. That's what most critics don't like about GPL - they can't take something "free" and create their own paid/closed version from it.

It's just a bad licence wrapped up in far too much shouty politics.

No, it's a good licence for some purposes - but I agree completely about the shouty politics.

[Alan Brown]

Re: It's the same old story with Linux - it's just one more thing

In the case of ZFS, Sun's CDDL was deliberately and explicitly written to be fundamentally incompatible with GPL

That's not the fault of GPL and for CDDL incompatibility is the desired feature, not a bug

[An_Old_Dog]

Playing nice with other recognized "open source" licenses

The reason the GPL doesn't "play nice" with other "recognized open source licenses" is that the GPL uses the structures and words needed to prevent people, or companies, from going Ferengi with your code ("What's mine is mine. What's yours is mine. What was yours, and is now mine, is no longer yours 'cause I did some legal tricksiness which your software license did not prevent, and you have no legal recourse. Ha-ha!").

This isn't a childrens' playground. It's the world of law, lawyers, and courts. The GPLs certainly aren't perfect, but they seem to best achieve effective protection against Ferengi-minded people and companies.

[VoiceOfTruth]

Re: It's the same old story with Linux - it's just one more thing

-> What GPL was designed to do was to prevent someone (I'm not naming any names, but certain large companies have been proved to do this) taking all the work put in to create software from being incorporated into their product, and then possibly IP protected via patent or copyright. It's this appropriation that GPL is to prevent.

Yes, I know what the GPL is meant to do. But look at where Linux is now. Much of the kernel development is in the hands of RedHat (IBM). The so-called Linux community often contribute new sets of shiny icons and desktop wallpaper. A RedHat licence cost more than a Solaris licence the last time I checked. The fact that I can get the source code at the same time is meaningless.

[Anonymous Coward]

@VoiceOfTruth - Re: It's the same old story with Linux - it's just one more thing

You're so confused, you should not wander alone here on this forum.

You're paying for RedHat/IBM brand, support and non-GPL software products but their Linux is free via the source code as GPL intended. To you, the source code is meaningless but not for these guys:

https://thenewstack.io/where-to-turn-for-a-centos-replacement-heres-5-solid-linux-distros-to-check-out/

[JamesTGrant]

Re: It's the same old story with Linux - it's just one more thing

I don’t understand why you post nonsense here - your GPL understanding is way off and the basic facts about Linux kernel commits by number of commits or lines changed is garbage (and very easy to verify). For example; for kernel version 5.10 RedHat contribution by change sets was 5.7% of total change, and by lines changed 3.9%, just higher than ‘Code Aurora Forum’. Even if you add IBM and RedHat together, the combined contribution is so far away from majority as to be a bonkers statement. Trend the contribution by vendor upto kernel version 5.17 and you see lumps per vendor but no one vendor contribution greater than 10%. Perhaps you’re thinking about systemd, which is not kernel? Please don’t make El Reg comments section less fun, coz that’s what you’re doing with highly agendaed (that’s a word now) bad facts posts.

[katrinab]

Re: It's the same old story with Linux - it's just one more thing

Yes, Linux is a kernel.

File system drivers are one of the things that kernels do, so you are running it in Linux.

[Batmensch]

Re: It's the same old story with Linux - it's just one more thing

I truly don’t see the point to this comment. What are you saying? Linux is bad? GNU is bad? What?

[NewModelArmy]

Re: It's the same old story with Linux - it's just one more thing

I ran a check on whether Windows can read Ext3 for example. The top searches all indicate you need third party software for Windows to read Ext3.

So, someone has contributed to Linux to allow it to read (at least) another file system which is proprietary (NTFS).

Microsoft does not natively support Linux Ext2/Ext3 etc.

Microsoft cares only about itself, and will only provide capabilities that it sees as being financially beneficial to them ?

The Linux project is an inclusive project - all are welcome. Microsoft are only interested in "things" if there is some money in it, for themselves.

Linux is great. There are many distros/flavours, so it is open and offers choice to the people. Microsoft does not, and actively deters openness.

[Stuart Castle]

Re: It's the same old story with Linux - it's just one more thing

Re "It's always been this way. It was Word format documents. It was SMB/CIFS. It's NTFS (and has been forever). It will be something else tomorrow.

Anyone seeing the response to the release of the Paragon code should take note. It was a thoroughly unwelcome attitude. Miserable ungrateful sods who can't do the job themselves but who think the world owes Linux something chimed in, and they repeated some of that here in a previous article. Some people even complained that because the code was once (and a version still is) closed source, it was bad. As I wrote - miserable ungrateful sods."

Sadly, some people do have the attitude that Linux is good, therefore everyone should bow to it, and Microsoft is bad, therefore everything they do is only worthy of scorn.

My attitude? Whatever OS you favour, all OSes have one thing in common.. They are tools. They either fit your needs or they don't. Your needs may not be the same as another person.

People need to be aware that just because someone uses a particular OS, they aren't necessarily better or worse than you, they are just different. I know experienced system admins who are good at their job, and entirely use Windows. I know experienced Unix admins who, given a choice, use macOS because it has a decent BSD implementation underpinning it, with a UI that is (in their opnion) better than anything offered by any *nix. I know others who use Linux almost exclusively, only touching Windows if required to,

They've all get their own preferences, for their own reasons. That doesn't mean that (say) the macOS users are better than the Windows users, or vice-versa.

Hell, I've been heavily downvoted on here because I had a project at work that if I'd been allowed to proceed, would have resulting in me managing hundreds of Raspberry Pis, and I made the mistake of asking if there was an enterprise level management system that enabled me to manage Pi OS on the the Pis in much the same way that System Center allows you to manage hundreds of PCs..

And yes, I work for a company owning thousands of PCs. The project, had I got funding to push it forward, would have required hundreds of Pis. Managing each one without the aid of a management server would never have been practical.

[VoiceOfTruth]

Re: It's the same old story with Linux - it's just one more thing

-> Sadly, some people do have the attitude that Linux is good, therefore everyone should bow to it, and Microsoft is bad, therefore everything they do is only worthy of scorn.

I have seen that time and time again here in The Reg. Alas. Four legs Linux good, two legs Microsoft baaad.

[NeilPost]

Re: It's the same old story with Linux - it's just one more thing

As a point of clarification… very few people in Iraq (and Afghanistan) were killed by coalition forces. The majority of deaths were down to post initial invasion/victory Moslem on Moslem Shia/Sunni sectarian violence, fiefdom building and religious nut job Mullah’s with their own private militias and outside interference by Regional Powers (Iran/Saudi).

Sorry if that’s inconvenient to your agenda.

That being said the lack of any post inevitable victory nation building strategy will be Blair and Bush’s fault. The lack of putting Hussain down post Gulf-War #1 is also under scrutiny.

[devin3782]

I'd welcome NTFS being open sourced and we've long been in need of a universal journalised file system that isn't the corrupt-o-nastiness of FAT, NTFS could fill that void especially as I don't see EXT4 being included in Windows.

What we really need is ZFS with a Linux compatible open source licence (which it currently doesn't) and BTRFS up to speed as its still a little too experimental in places.

[A Non e-mouse]

I don't see EXT4 being included in Windows.

devblogs.microsoft.com/commandline/access-linux-filesystems-in-windows-and-wsl-2/

[devin3782]

Oh! very interesting thanks I stand corrected, I'm also so pleased they've made mounting them needless complicated well done Microsoft.

[Alumoi]

Windows Insider only. Wake me up when it comes to the (less) buggy one.

Before you all start waving pitchforks and torches, wsl --mount doesn't work on Windows 10 Enterprise LTSC 21H2.

Still, an upvote for reminding us.

[doublelayer]

If they're going to add it in, it has to be tested first. It's a new addition, just as the read/write NTFS support is new in Linux. It's going through the same system they always use for releasing new features. Why is it so bad that it isn't released yet? I'm having trouble identifying what you would want instead, as if they simply sent it out from dev to full release, I bet you would have several (correct) complaints about adequate testing.

[Alumoi]

If they're going to add it in, it has to be tested first.

Like they're testing their updates?

It's a new addition, just as the read/write NTFS support is new in Linux.

NTFS-3G anyone? New as in only 15 years old?

[Zippy´s Sausage Factory]

If they're going to add it in, it has to be tested first. It's a new addition, just as the read/write NTFS support is new in Linux. It's going through the same system they always use for releasing new features. Why is it so bad that it isn't released yet? I'm having trouble identifying what you would want instead, as if they simply sent it out from dev to full release, I bet you would have several (correct) complaints about adequate testing.

Ah yes, Micros~1's testing system that seems to consist of

(a) release

(b) get bug reports

(c) close bug reports with snarky comments as to how users are using it in a way for which it wasn't intended

(d) sit down and have a cup of coffee with Satan.

[Tmanok]

Don't forget:

F) point and laugh when the releases causes an especially ridiculous problem like the Domain Controller reboot loops from earlier this year or the recurring routing packet BSOD bug that has plagued W10 at least 3 times after being patched and then rebroken.

[VoiceOfTruth]

-> What we really need is ZFS with a Linux compatible open source licence (which it currently doesn't)

So change the GPL to something which accommodates ZFS. The GPL is a horrible licence. It is the gold standard of non-interoperability, and is designed exactly as such.

[RichardBarrell]

Not to get into too much of an argument here, but the gold standard for non-interoperability in software licensing is the AGPL.

[VoiceOfTruth]

Could be, actually.

[bombastic bob]

Fortunately Linux is still GPLv2 last I checked.

There is also nothing stopping anyone from including ZFS as a build-from-source install. I do that with Linux drivers sometimes. Debian derivatives have the 'module-assistant' package for that purpose.

[georgezilla]

Bullshit.

[Chris 15]

False

Just because it precludes free lunches by people taking and redstributing it without granting the users of the redistribution the same rights as the person redistributing it does not make it a problem for interoperability. you can have your cake and it it with the gpl, just no cherry picking. :p

[nijam]

> So change the GPL to something which accommodates ZFS.

Why? Remember that Sun licensed ZFS under a licence that they deliberately created to be incompatible with GPL. That is absolutely not the fault of GPL.

[VoiceOfTruth]

It was Sun's code. They can licence it how they wish. They did not owe Linux a bean.

[Anonymous Coward]

@devin3782 - How about

OpenZFS ?

[devin3782]

Re: @devin3782 - How about

OpenZFS uses CDDL which to my knowledge incompatible with GPL, I'm sure I read somewhere deliberately so. When I mention ZFS I did of course mean OpenZFS considering Oracle's version isn't open source and probably been killed off by now, but its Oracle of whom I don't give a dingo's kidney for.

Of course I also don't give a damn about the licence, I'm using OpenZFS it on my Debian server as VM pool spread across two nvme drives and its thoroughly lovely.

[emfiliane]

I used to be the world's biggest fan of btrfs, but gave up on it thanks to its utterly glacial pace of work and unwillingness of any maintainer to tackle difficult pain points. And every maintainer has been very temporary, because most of what's left are only difficult pain points, so there's a flurry of activity for a couple of weeks or months, then it goes on life support again until someone else gets into it.

The situation isn't that much different from Linux NTFS's maintainer vanishing into thin air, except at least that project was completed.

Advanced filesystems are so painfully complex that it seems the only way to get them built is a paid team dedicated just to that, like Sun did for ZFS. Red Hat and Canonical aren't nearly excited enough about btrfs to dedicate that much capital to it, the way they have for improving XFS.

[VoiceOfTruth]

-> Advanced filesystems are so painfully complex that it seems the only way to get them built is a paid team dedicated just to that, like Sun did for ZFS.

Years ago I read a report or a story about IBM and file systems. I'm paraphrasing from memory but it's along these lines: there's only a few people in the world who can make a great file system, we know as we've tried to hire all of them.

[Peter Gathercole]

I don't know which filesystem they were talking about, but jfs2 built on top of lvm (on AIX) is an absolutely great combination, and apart from some performance problems on modern large disks has stood the test of time really well. The open jfs2 is actually a re-implementation, by IBM, from their OS/2 code base.

When you consider when it was written, and in fact that the predecessor (jfs) pre-dates pretty much every other scalable filesystem (first appearing in 1989 in demo AIX 3.1 systems), I think that IBM has some skin in the game.

And then you have GPFS, which is great for huge, distributed, high performance filesystems. OK, it's proprietary, and IBM makes a pretty penny licensing it, but its pretty good.

[Anonymous Coward]

Surely nobody uses NTFS for anything more than a system disk for Windows?

All the actual data should be on a network drive, which is most likely a Linux box running ext4 with a samba share.

Even enthusiasts who run dual-boot PCs at home have a NAS these days.

Is there any use case for ntfs support in Linux apart from system rescue CDs?

[VoiceOfTruth]

-> All the actual data should be on a network drive, which is most likely a Linux box running ext4 with a samba share.

You live in a very closeted world. We do use network shares, but we don't use EXT4 for use with Windows machines. There is no benefit at all to doing so.

[georgezilla]

" ... There is no benefit at all to doing so. ... "

Neither is there in NTFS.

That's why I haven't used it in 20 years.

[bombastic bob]

Is there any use case for ntfs support in Linux apart from system rescue CDs?

I remember this being brought up in an El Reg article about Paragon adding their NTFS driver into the Linux kernel. As I recall, dual-booting and using existing data were two good reasons.

As for me I do not dual boot any more, VirtualBox doing what I need for those cases. But others still dual boot.

(I actually had a 'quad boot' set up once with MS DOS, Win2k, Win 95, and FreeBSD 4.11 - it was a little klunky but it worked - on a 486 DX2 with some pathetic amount of RAM on it)

[CrazyOldCatMan]

As I recall, dual-booting and using existing data were two good reasons

Indeed. In the dim and distant past we didn't have reliable hypervisors so the only way to do a multi-OS setup was to dual boot.

Where I was doing that sort of thing I usually had a Windows partition, a linux partition and third FAT32 partition for stuff I wanted both to be able to get to. Which wasted a fair amount of space.

Nowadays I have Parallels on my Mac, Hyper-V on my work Windows laptop and KVM on my virtualisation servers upstairs (yay for Proxmox!) so really, really don't have need for multi-OS booting.

Obviously my requirement isn't universal. And the entertainment unit in the car only accepts USB sticks formatted with FAT32.. (Not that I think it'll ever be able to read APFS - all my music is on my Mac!)

[Anonymous Coward]

@bombastic bob - Honestly, Bob,

how many times have you seen dual-booting on a production server ?

The quad boot you're mentioning is remarkable but can you put this into production in a corporate environment today ? I'm not excluding this use case but I don't think it is that widespread for someone to invest resources in it.

On to the other hand, data recovery is a valid use case, especially if we take into account the scarcity of tools provided by Microsoft for this purpose.

[bombastic bob]

Re: @bombastic bob - Honestly, Bob,

on a server you might choose to mount an NTFS formatted drive with legacy data on it, possibly for use by a VM or application running under Wine, or who knows what unforseeable reason. Edge cases, but still...

and Linux is not just used on servers, FYI.

[Toe Knee]

Win2k on a 486? That must have given you plenty of time to get other things done while you waited on it to do anything. At least it was a DX!

[bombastic bob]

amazingly older versions of Windows NT 4 and 2k worked pretty well on 486 and Pentium I. I noticed a significant reduction in efficiently with Win2k3 though, but i had Win 2k server running on a Pentium I 133Mhz without any serious performance bottlenecks, doing basic network and file share things. When i loaded win2k3 server on the same platform, it was klunky and stuttered a lot... (even though it was only for testing it was so pathetic it was the last windows server version i ever installed on ANYTHING)

[doublelayer]

"Is there any use case for ntfs support in Linux apart from system rescue CDs?"

Of course not. Nobody needs more than one filesystem anyway. Looking at my /proc/filesystems, I see 42 supported ones in here. Let's drop 41 of them. Surely nobody will notice, right?

There are a lot of systems using a lot of weird configs. You can't find a single solution to them all. The list I've brought up is from a server with a basic config and where I haven't added any additional filesystem modules, and some of the systems in there have never been used (I don't create firmware images on it, so the squashfs support is not needed here). It is still used by many, including me on other systems, so it stays in. When something doesn't get maintained, it will eventually be cut from the kernel. There's a reason that, far from dropping the NTFS support previously present, they've added a better version. People have a use for it.

[doublelayer]

One additional use that has already been mentioned here is multiplatform storage. I would like this, too. I have machines running a variety of OSes to which I'd like to attach a storage device. Connecting all of them to a central server and doing all the file access over the network is slow and requires configuration, whereas connecting a USB cable to a hard drive is much easier. A filesystem that can be reliably read and written no matter what system it's being used on would be useful. It's not just technical users; even the nontechnical users tend to have Linux running on embedded devices of various kinds to which they attach portable storage. I know this because they've frequently called me for help when those devices failed to recognize the storage they attached and I had to talk them through backing up the data and changing the filesystem in use so the device would accept it.

Unfortunately, the one that you can pretty much count on being supported is FAT32, which despite its ubiquity, is not a very good file system. A lot of things still use it because of its wide support, which results in file size limits, reformatting devices when it's the only option, etc. The system to replace it doesn't have to be NTFS, but it has to be at least a little like it and for now, NTFS has read support nearly everywhere so it's further along.

[Piro]

Newsflash, NTFS is actually fine, and of course it's widely used.

In the real world, countless companies have countless files stored on NTFS.

[Matej101]

Needed if you want to dual boot Windows and have access to the same data from both OSs.

Yeah it's a weird edge case, but for example pretty much ALL of the widely used 2D and 3D CAD packages are Windows only. Solidworks even has a handy feature that prevents you from running it in a VM unless you have a special (more expensive) license.

Why use Linux at all then? Well I need things like Socketcan sometimes.

[Matthew 25]

'Even enthusiasts who run dual-boot PCs at home have a NAS these days.'

1. Not all of them

2. Some off the shelf NAS use NTFS

[theOtherJT]

re: which is most likely a Linux box running ext4 with a samba share.

...you're not thinking about this at scale. Full Windows shop it's probably ReFS. Under Linux I'd expect it to be something like ZFS or if we're really talking scale: gluster, or ceph, or ocfs2, or gpfs or one of the other hundreds of network distributed filesystems. I've seen far too many very ugly things happen to ext4 on LVM when people thought that was sufficient for heavily used network share.

[SloppyJesse]

All the actual data should be on a network drive, which is most likely a Linux box running ext4 FreeBSD box running zfs with a samba share.

FTFY

[b0llchit]

Microsoft may have embraced the penguin, but it still thinks using Windows 11 as an advertising platform is a great idea.

Maybe they will bother to "fix" the NTFS driver if they can add a little advertising code in the driver? Every file you open starts with an ad. Or images are transparent ads and change over time. I'm sure they can find a nice advertising touch. Maybe a new and fresh ad every time you open the file? That would be the best of both worlds, wouldn't it?

_{I'd be damned and burned at the stake if proven right on this one}

[ThatOne]

Well, to play the devil's advocate, Microsoft is most likely just having one of their usual marketing-induced psychotic events.

Remember: At some point somebody realized that "oh crap some people are making money with smartphones", and everything Microsoft touched became a phone (see Win8). Then somebody realized that the big players out there were into ads, and Microsoft harum-scarum rushed to jump onto that bandwagon. It will most likely pass as soon as they discover their next get-rich-quick scheme.

It's bandwagons all the way down with Microsoft...

[bombastic bob]

when your company revenue has been based on the ability to sell ice cubes to indigenous peoples in frozen wastelands, it's just more of the same kinds of marketeering.

[A Non e-mouse]

Microsoft Bashing

Oh please, can we all stop the childish "Why does anyone use Microsoft software?" In your little fantasy world you may be able to survive as a "pure blood" but the rest of us live in the real world and have to deal with Microsoft, Oracle, IBM, et al. I'm not saying their products (or the companies for that matter) are wonderful: I swear at them on a daily basis. But my employer has decided that software X is what it wants and either I help support it or I look for another job. I doubt there are many IT jobs where you don't touch Microsoft (or Oracle or IBM) products.

[VoiceOfTruth]

Re: Microsoft Bashing

-> the rest of us live in the real world and have to deal with Microsoft, Oracle, IBM

Yes. This is why I am bit bored of Linux people in particular, but *some* open source advocates saying 'just use Linux or open source' when it does not meet our needs. I'm a fan of Linux in general, but not of the umpteen distributions which is just wasted effort. I'm a fan of open source, but sometimes it does not meet our needs.

I'm also very wary of the false information that open source software is more secure than closed source software because more eyes are supposedly looking at it.

[Adair]

Re: Microsoft Bashing

'I'm a fan of Linux in general,...' – interesting assertion, as over and over again you demonstrate a refusal or inability to understand what 'open source' is all about. Happily, you are free to use it, or not, and, if it really doesn't suit your needs, to find something that does—whining about it is purely optional.

[VoiceOfTruth]

Re: Microsoft Bashing

Perhaps I understand it better than you, and it is you and people like you who are wearing the blinkers.

[georgezilla]

Re: Microsoft Bashing

Lol.

Okay if you say so.

But I doubt it.

[Runkel]

Re: Microsoft Bashing

Didn't you just recently demonstrate that you tend to confuse distributions with desktop environments?

[VoiceOfTruth]

Re: Microsoft Bashing

No confusion at all on my part as there two old/dead distributions being brought back to life.

[Runkel]

Re: Microsoft Bashing

Not distributions. Desktop environments.

Distributions are e. g. Debian or Nix, ideally embodying principles of OS management.

Desktop environments embody principles of UX in the best cases, which doesn't mean that all of them are actually doing it.

That you apparently can't tell one from the other doesn't speak for your expertise.

[georgezilla]

Re: Microsoft Bashing

" ... I'm also very wary of the false information ... "

And you have actual credible proof to back up that statement?

Please provide it.

I'll wait .....

No.

Actually I wont.

[VoiceOfTruth]

Re: Microsoft Bashing

I will tell you why it is false information, and you can either accept or ignore it because it is the truth.

Firstly there is the so-called Linux community (and the open-source community). Most of the people in these communities are users, not programmers. That's fine, I accept that. I don't expect every Linux user to be a programmer. Next, of those programmers not all of them are good programmers. Again, that's fine. I don't expect them to be. These people generally do not find important bugs. They might find typos in documentation or in translations and so on, or they might point out bugs. That's fine, that is all good and essential in my view.

Now you start to get to the *good* (or at least better) programmers. What are they working on? That depends in some cases on who pays them, what they are interested in, and what they are knowledgeable about. A programmer might be a Python wizard who works on installation GUIs. Fine. But they likely won't be looking for or fixing bugs in OpenSSL. Or Log4J. Or pfexec. Or...

Who is working on these things, who is looking for these bugs and security holes which are arguably far more serious? Sometimes bugs may be very tricky to spot, and to do that you need not 'many eyes which are not very good', but 'few eyes which are very sharp'.

I've been in this game for a long time and I've seen and heard a lot. I have heard many many times over the years 'the community this and the community that'. This is often from people who may be part of the Linux or open-source community, but who themselves are not capable of looking for or fixing bugs. They just aren't.

It is false information to claim otherwise.

[Martipar]

Re: Microsoft Bashing

I agree in full. I have used Linux on and off for 20 years or so and every time I think "Linux should be able to do that." I either find it can if I use 3 different programs or it can't unless I install WINE and run the program I want.

IT managers also know this, they know they can just buy some hardware, some Microsoft OS and it'll just work and it'll be compatible with every other company as they likely use Microsoft software too. If Linux worked and worked well it would be adopted but there's so much fragmentation that outside of the server side of things it's not worth the hassle. In 2012 I decided to run Linux (and only Linux) on my PC, normally I dual booted, ran a live system or had it on my secondary PC never my main one. I decided to buy a PCI wireless card, on Windows that's easy, look at the specs and the price, work out which one offers the best VFM and buy the one I want. On Linux it's find one I like, then spend ages trawling through technical documents to find out what chipset it uses then finding out if that chipset is compatible with Linux and that's just one piece of hardware.

I still have trouble with one of my laptops showing a black screen if I want to run Linux live on it unless I switch to the safe graphics mode because Linux does not do what Windows does and show a reduced graphics mode automatically.

I have a laptop connected to my hi-fi, all it has to do is store my music. play them and show visualisations I have used Winamp for 20+ years because it's awesome, and the media library is perfect I spent hours, easily 5, trying out every Winamp alternative on Linux and other media players only to find the only one that had a library I liked Quod Libet - didn't support visualisations, i then spent more hours compiling a visualisation program from source because Linux devs are allergic to compiled binaries, listing where dependencies can be found and the make command can't list all the missing dependencies in one go rather it lists each one as it comes across them.

Needless to say the visitation program didn't work, it was at this pint about 4am and I took a deep breath, reached for my XP CD and wiped the drive and regretted spending 30 minutes transferring my music to the HDD.

I could go on but I won't for fear of writing a novel about the whole thing, I like the idea of Linux but often in practice it does not do what i want and all I want it to do is function properly. One day i'll find a reason to use it over Windows but for the last decade I've happily bought Windows because i'd rather buy Windows than have Linux for free.

[georgezilla]

Re: Microsoft Bashing

" ... In your little fantasy world ... "

Except for the most part, the real world uses Linux for almost EVERYTHING.

You know, things that Windows is shit at.

And most people that I know ONLY use it for one of two reasons ......

1) They have used it for so long that they actually are stuck with it.

2) They are ignorant of the fact that there is something else that they could use.

And most of them could absolutely use Linux with no problem.

( Well except for one guy. He still has trouble answering his smart phone. )

" ... (or Oracle or IBM) products. ... )

Hint: Both of them have and support Linux distros.

[DeathSquid]

Re: Microsoft Bashing

I don't know what your "real world" is like, but over the 20 years I've worked for 3 fast growing Fortune 500 organisations and not had to touch microsoft crap personally. Over the last six or seven years, I basically haven't seen windows used by anyone I work with. It's all Linux, Macs and mobile OSes.

The 90s are well and truly over. The future is the cloud and mobile devices.

[42656e4d203239]

Re: Microsoft Bashing

>>It's all Linux, Macs and mobile OSes.

Which is to say (for the sake of argument) Linux, (these days) Unix and Linux (Android and ChromeOS)+ iPhoneOS which is Unix derived...

Linux/Unix on the desktop is already here; it's just not here in the way one might expect.. We have more Chromebooks than PCs here at work - so, if that is true elsewhere, the days of Windows are, surely, numbered?

[DeathSquid]

Re: Microsoft Bashing

Indeed. What cracks me up are the desperate downvotes by people who are stuck in the past for one reason or another. They puff out their chests, harrumph, and declare how things are in the "real world". I've been in this game for 40 years, and every tech cycle I see those who don't want to or can't change getting left behind. It's a cruel business.

[bombastic bob]

"but if 2022 has any lessons for us, it's that we can't have good things"

sadly, the best. quote. evar.

[Steve Davies 3]

Re: "but if 2022 has any lessons for us, it's that we can't have good things"

That should be...

but if 2022 has any lessons for us, it's that we can't have good things because some [redacted] politician or cable news anchor wants to ban it and stop us from having fun.

Wanna have fun? Watch out Fun is an endangered activity and could damage your health, wealth and well being.

[Yours... The Ministry of Truth]

[Wade Burchette]

Microsoft should move beyond NTFS

I think Microsoft should move beyond NTFS. And I am not talking about their ReFS. What we really need is for the Apple, Linux, and Microsoft people to all get together and come up with a universal, open-source, free, and interoperable file system. No more proprietary file systems. This would benefit everyone, including Apple and Microsoft. But I am certain that because everyone would benefit, a universal file system will never happen.

[bombastic bob]

Re: Microsoft should move beyond NTFS

How about ZFS ?

[theOtherJT]

Re: Microsoft should move beyond NTFS

Much as I love ZFS I'm not sure it's a great choice for the standard use case "One disk - one filesystem" setup. Been a while since I did those benchmarks mind - SSDs were rare and expensive last time I did. Maybe the fairly shocking performance hit you got on a single spinning disk zvol compared to the same disk on XFS don't apply any more... If that's then case then yeah, camp ZFS. Sign me up.

[VoiceOfTruth]

Re: Microsoft should move beyond NTFS

Which licence would it have? If your answer is GPL, forget it. There is no way that virus will find its way into Microsoft or Apple. If the answer is BSD, then yes that is possible.

[georgezilla]

Re: Microsoft should move beyond NTFS

" ... There is no way that virus ... "

Interesting.

Many, many of us feel that way about Windows.

And many of then still use Windows in spite of feeling that way.

[VoiceOfTruth]

Re: Microsoft should move beyond NTFS

Does Microsoft compel you to write code in a licence of its choice? Yes or no? Back in the naughty corner with you.

[Anonymous Coward]

@VoiceOfTruth - Re: Microsoft should move beyond NTFS

Just try to include some Microsoft code in your software and come back to tell us how free you are in choosing your license. Pick some serious code like file system or networking and see if a simple attribution clause will suffice.

Let's not be silly, shall we ?

[SImon Hobson]

Re: Microsoft should move beyond NTFS

Neither does the GPL ... now go and think for a bit until you understand why,

If you think the answer is "but it's going into the Linux kernel, it has to be GPL", then you don't understand licensing.

[ThatOne]

Re: Microsoft should move beyond NTFS

> because everyone would benefit, a universal file system will never happen

Close: It's actually "because only the clients would benefit, a universal file system will never happen".

They have no reason to make one, because commercially it would be harmful. It would remove barriers used to prevent clients from using the competition's products instead of yours. You'd have to fight the competition on merit, which means spending money you could had put in your pocket. And all this, what for, just to feel good about yourself? You'd feel even better with a couple millions more in your profit margin...

[A Non e-mouse]

Re: Microsoft should move beyond NTFS

I think you'll find the design of a filesystem is closely aligned to the primary operating system it's going to work with. If you try to design a filesystem that works with Linux, Windows, Apple, etc, you'll either end up with a very basic filesystem (e.g. something looking a lot like FAT) or something so hideously complicated that no-one will use because it's too slow, bug ridden and complicated to debug.

And I'm not even getting to the differening workloads/use cases placed on a filesystem.

We all know which XKCD we're thinking of here.

[An_Old_Dog]

Designed "universal" standards

Everyone who thinks standards designed to be "universal" are good ideas need to have their heads adjusted with a clue-by-four.

These standards are created by people who long to architect huge, great monuments to the awesomeness which they believe themselves to be. (USB, XML, PulseAudio, and Systemd are examples.) Adding large corporations -- each with their own ulterior motives -- into the mix makes things worse.

Good "universal" standards typically are simple things which work well, and which never were intended to be "universal". The Centronics parallel printer interface is one such.

[bombastic bob]

Re: Designed "universal" standards

well, universal standard time is a good idea, if you like air and sea navigation to work, and trains to share track without crashing.

Then there are standard sizes for electrical connectors, screw sizes, peripheral sizes (like disk drives), memory connectors, CPU sockets, IC footprints, yotta yotta yotta.

Then there are things like USB (why is it bad?), HDMI, CD/DVD/BluRay formats, WiFi modulation, even broadcast radio - the list goes on and on. Without universal standards NONE of these would even work.

I suspect that there may have been at least SOME "in our best interests" creep in their development, but they are universal and agreed to and adhered to enough that there is overall benefit from the standardization.

[VoiceOfTruth]

Re: Designed "universal" standards

-> Then there are things like USB (why is it bad?), HDMI, CD/DVD/BluRay formats, WiFi modulation, even broadcast radio - the list goes on and on. Without universal standards NONE of these would even work.

Yes. And goodbye to the internet with its built upon universal standards in the RFCs.

[georgezilla]

Re: Designed "universal" standards

Careful.

You are going to confuse them with fact and reality.

And they tend to get nasty when confused.

[IGotOut]

Re: Designed "universal" standards

Universal standards such a Litres, Pints and US pint?

CDs. Please, which "book" standard are you referring to?

Ever world I the International Standards of telecoms? Have fun getting the standard g711u talk directly to g729a/b with a converter?

Connected many plastic water pipers to copper ones?

The standards are great, but in reality you usually have to bodge something in between.

[J. Cook]

Re: Designed "universal" standards

The phrase "I love standards, there are so many to choose from!" comes to mind.

[VoiceOfTruth]

Re: Designed "universal" standards

-> Universal standards such a Litres, Pints and US pint?

Our pints are bigger.

[Paul Crawford]

Re: Designed "universal" standards

Then there are things like USB (why is it bad?)

Many reasons:

Stupid naming of different speeds

Early versions had no useful interrupt/DMA capability

Excess complexity compared to RS232 for most jobs

Implementations made the "oh I have found X device, lets run it! Oh bugger that was a trick, we are screwed"

[An_Old_Dog]

Re: Designed "universal" standards

@bombastic bob: universal standards (vs ordinary standards) strongly attract "monument-builders", who create exceedingly complex standards, with exceedingly-vast functionality. Exceedingly-vast funtionalities require exceedingly-complex hardware and software to implement them.

Excess complexity is where mistakes-slash-security holes breed.

[Paul Crawford]

Re: Designed "universal" standards

Indeed. While developing a USB device many years ago I managed not only to crash the XP host with my microcontroller's code, but also to wipe the boot disk MBR FFS!

Had to get the local folks to re-image that machine to continue (I was not permitted any admin rights on their test network, etc).

[georgezilla]

Re: Microsoft should move beyond NTFS

An attempt to do just that has been made.

The Open Document format/standard.

Microsoft actually adopted it.

And then imminently tried to fuck it up and kill it.

[SImon Hobson]

Re: Microsoft should move beyond NTFS

Err, actually Microsoft never adopted Open Document Format.

They may have added some token and broken support for it, but they never ever "adopted" it. What they did do was to spend a lot of effort and money pushing their own closed and not at all open "open" standard though ISO standardisation so they could tick the box on tendering documents.

But be under no impression that they have in any way "adopted" any form of open standard for office documents.

[Nugry Horace]

Re: Microsoft should move beyond NTFS

There's even one out there called Universal Disk Format (it's what DVDs use). I remember seeing recommendations to format your USB sticks as UDF to get around FAT limitations. Never caught on because operating system support for using UDF on anything other than a DVD suffered from all sorts of weird limitations.

[Anonymous Coward]

@Wade Burchette - Re: Microsoft should move beyond NTFS

You're a little bit un-capitalist here (I wouldn't go as far as calling it a communist view). The purpose is not to benefit everyone, it is to benefit one more than the rest of them. Remember, capitalism is about making money not about spreading benefits.

[Piro]

Re: Microsoft should move beyond NTFS

Open sourcing NTFS is a far better solution than them just inventing another new barely tested system.

ReFS has so many weird quirks and caveats after all this time.

[Julz]

Live

By open source, die by open source. Useful/vital free open source software being maintained by a single or very small group of persons out of the goodness of their hearts is the norm. It has always be thus.

[georgezilla]

Re: Live

Lol.

Tell that to any industry that uses opensource software, for ANYTHING other then desktops.

And many that actually do use it on the desktop.

[Anonymous Coward]

Re: Live

They just bull brittle systems because of the greed. It's the "the lone developer from Nebraska" in the xkcd strip. There was the news that PHP was actually maintained by two people and one was leaving, IIRC. The OpenSSL debacle for lack of maintainers. One day something bad may happen, it's just waiting to happen.

This is a failure of the open source model - too many projects don't know how to pay developers so they lack maintainers.

Why Microsoft should fill this void and release an open source NTFS driver for Linux? It really has no reason do to so. The "open source community" should find people to maintain the Paragon code, if that's so badly needed by the community itself, isn't it the way open source should work? Maybe paying them, or finding a monk in Nebraska wishing to maintain it for free...

[VoiceOfTruth]

Re: Live

-> One day something bad may happen, it's just waiting to happen.

We saw it with Log4J. And how some uninformed people howled.

-> This is a failure of the open source model - too many projects don't know how to pay developers so they lack maintainers.

Not only that, I would add a lot of people do not want to pay. Money talks. Like buttons don't.

[An_Old_Dog]

Failure of the Closed-Source Model

... is what happens when companies making a product you depend on go out of business, or "change stategic direction" and consequently quit making/supporting the product you use, and give you the following options:

(1) No upgrade path whatsoever (no security patches, either);

(2) Upgrade to a product which lacks features and/or the speed you need; and/or,

(3) Upgrade to a product which is not cost-effective for you.

I don't say open source will save the world, but I do say closed-source fans have to (if they're being intellectually-honest) acknowledge the disadvantages of closed-source.

[Ben 56]

Back in the day...

I was tasked as a junior dev going to one of these "Linux is evil" shows Microsoft did, showing TCO was indeed greater with Linux, especially staff training and pre existing integration (no surprise incumbent needs less work there then). It also mentioned patent breach exposure as a risk (and MS was threatening people, SUSE paid MS for patent protection).

However during one of the Q&A sessions I blurted out "why does the TCO not contain the an estimate for the amount of time needed to patch or fix CVEs?" of which Microsoft at that time had recently had a very large ring 0 worm shortly after the sessions started. I also asked if the cost of repairing said breaches was in there too. Petulant maybe as a junior dev, but I hated people being misled by lies, damn lies and statistics.

Unsurprisingly, the questions were glossed over and the answer not given, just re-iterating what was in the TCO summary of training costs etc.

I've never looked at MS the same again, and proof they still haven't changed is evident by them never letting go of their clasp of the old crown jewels.

IMHO you have as much chance of getting the NTFS driver as getting Windows >9x open sourced.

Having said that I own an XBox...

[Anonymous Coward]

Er, didn't a commentard suggest this exact thing 2 weeks ago ?

I know they did - it was me.

[Steve Channell]

nothing to do with OS/2

OS/2 had HPFS ("High Performance" File System). NTFS was purely an NT file-system.

Now that Windows has moved to ReFS, there's a good case for NTFS to be ported to Linux, but should also include DLL loader for PE files

[CrazyOldCatMan]

Re: nothing to do with OS/2

OS/2 had HPFS ("High Performance" File System). NTFS was purely an NT file-system.

Microsoft (at the time) sort of admitted that NTFS grew out of HPFS - it was very similar in structure at the start (and yes - putting an NTFS drive into an OS/2 server and tweaking the filesystem partition ID was a Bad Idea (TM). It didn't last long before getting hopelessly corrupted..)

They started off as cousins and grew further apart as HPFS grew then disappeared and NTFS changed with each iteration.

[Gnisho]

Re: nothing to do with OS/2

Shortly after the MS/IBM split on OS/2, word was the only difference between NTFS and HPFS was the filesystem header though they evolved apart quickly afterwards.

[93s]

Re: nothing to do with OS/2

The original 16-bit HPFS may have been a joint MSFT-IBM product, but Microsoft created HPFS386 (the 32-bit version with integrated SMB), and owned HPFS386 after the divorce. Microsoft has admitted that early versions of HPFS are based on NTFS (though they've never been entirely clear how much of that basis was 'conceptual' versus literal code identity). The low level structures of the original NTFS and HPFS are pretty similar.

[Anonymous Coward]

"seems as obsolete a concept as decisions about tape formats."

*looks at a DAT72 tape drive, which is at this very moment restoring a backup*

Yeah. Obsolete and nobody has to be concerned about it any more...

[cjcox]

What??

Microsoft teases Linux. They definitely do not love Linux.

If they loved Linux they would support their products on it. They don't. You say, "but what about Teams? SQL Server? Powershell? WSL?"

Actually, it's part of my point. Because in all of those cases where Microsoft has a temporal port or build or integration to Linux, it is very poorly supported. And their track record for long term support on Linux is horrendous. It's a closed source company, and they believe in their model (only). That is, you drive your workers to make your software better by giving them rewards/punishment. You will not find a lot of coding pride there, because it's just the wrong culture. Software that costs them money to make/support and nets them nothing, gets cut.

So, with regards to NTFS. Don't use it.

It's a primitive file system in many ways, it's designed for Windows. I say, leave it there.

Let's just leave Windows on its rapidly becoming deserted island. They don't want to leave their island, and they certainly do not want to be helped. They're happy there.

[VoiceOfTruth]

Re: What??

-> Let's just leave Windows on its rapidly becoming deserted island.

And he cast the runes and predicted the Year of the Linux Desktop is nigh, and when Windows 7 comes out everyone will move to Linux, and when Windows Vista comes out everyone will move to Linux, and when Windows 8 comes out everyone will move to Linux, and when Windows 10 comes out everyone will move to Linux, and when Windows 11 comes out everyone will move to Linux.

And later when he saw that his soothsaying was wrong every time he cast his runes into the abyss and said sod this. And he was happy with his Gnome desktop and its new text editor and some new desktop wallpapers.

[Anonymous Coward]

@cjcox - Re: What??

See what you've just done ? Did you really have to provoke the voice of truth when he was showing signs of a calm down ?

[NiteDragon]

Re: @cjcox - What??

I suspect they need to reply at almost 1:1 for linux related comments or a little bomb goes off. It's like a really tedious IT version of the film Speed.

[doublelayer]

Re: What??

Microsoft likes Linux as a developer and server platform, mostly from their Azure people, but they're not going to put all their effort into it, not when some people still pay for Windows and products that run on it. Teams on Linux might not be great, but I have a secret to tell you: Teams on Windows is ... also not great. Their developer-focused stuff is generally better, though it is still young compared to tools that started with Linux as a target.

Microsoft is never going to decide to be Red Hat. They'll do some of the things that Red Hat has done, but they'll do it when it suits their business. Stuff that makes Linux VMs in Azure more popular will get done. Stuff that attracts developers to coding that will also work well on Windows will get done. Writing software that runs on servers so that it can also be used by the millions of people using Linux servers will get done. Writing Office from scratch so it runs on Linux, when they already have a web version that will run and they know most Linux users are perfectly happy using LibreOffice instead, won't get done. This doesn't make them an adversary.

[Binraider]

Re: What??

Many platforms need access to more than one technology; e.g. data drives on a dual-booting desktop; I'd like to be able to (safely) read/write from both host OS.

In a corp environment there are loads of reasons to want to be able to read/write to and from NTFS from a non-windows host.

A properly maintained driver benefits many users; not least of which would include MS products like Azure. Given direction of MS desktop activity; Windows 12 or 13 on top of a Linux front end I would not at all be surprised to become a thing. Never say never - ever use Xenix? WIth it's weird and wonderful hybrid of forward and backslashes depending on what you were doing on what file system...

I don't know what the state is of MS kernel developers today but I suspect it is rather backward compared to the early days of NT where; whether you love it or hate it, MS did have access to some excellent programmers at the time. Drawing on the best resources out there makes sense for them.

MS don't need to support the driver themselves but they can release the detail needed for the community to do it.

[kurkosdr]

Microsoft will not open-source NTFS because:

1) It's a complicated set of code, which means they have to put effort into writing the Linux kernel driver

and

2) They will have to commit to a stable specification. Without committing to a stable specification, they can make changes to it such as expanding it etc and it's the job of Linux kernel maintainers to catch up.

Just use exFat for your external drives. NTFS was never meant for external drives anyway.

[Philip Storry]

I'm not convinced

I'm not convinced that Microsoft could - or should - do this. My reasoning requires a little knowledge of filesystems to understand...

Filesystems are composed of two things - the filesystem layout; and the software stack.

The filesystem layout is the actual bit written to persistent storage - everything from header blocks to identify versions and features, disk maps (if used), directories/inodes and so forth.

The software stack is the actual code. And it's complicated. It's not just a bit of code that reads and writes the filesystem layout, it also has to interact with a variety of other operating subsystems. Memory management, buffered I/O, block storage drivers, caching systems... Security can be handled at a higher level, so I'm not adding it to that list, but it's also a possible complication.

Open-sourcing the current stack would be pretty useless, as it's Windows-specific. Who knows what oddities lie within its internal APIs to handle file locking or caching systems that are Windows specific?

Because of this, it would be a fairly major job to write and then release an open-source driver.

Hence me not being convinced.

What they could do, though, is release two documents that would help immensely. Documentation for the filesystem layout, and a technical note confirming the order of operations would both be very helpful to those in the Linux community who want to write and maintain an NTFS driver.

[Marty McFly]

Might want to check....

Back in the late 1990's, Veritas signed a licensing agreement with Microsoft Corporation to provide storage management software for the Windows NT operating system.

I don't know what parts of NTFS are owned by Microsoft and what parts come from Veritas. It is possible Microsoft may not be able to open-source NTFS even if they want to.

[James Turner]

Re: Might want to check....

The later lawsuit by Symantec (who bought Veritas) was over their Volume Manager. That would tend to sit underneath the file systems rather than being a part of it.

I suspect a more likely source of encumbrance would be IBM depending on how the OS/2 and NT divorce was sorted out.

[martinusher]

NTFS -- chasing the rainbow

Microsoft has used unnecessary complexity as a marketing tool since the beginning - it impedes competition and the myriad small bugs and gotchas makes it difficult to build compatible components. Understanding this explains why reverse engineering NTFS is chasing a moving target -- as soon as you get close they'll move the goalposts a bit, just enough to produce subtle, but important, bugs in your code.

I don't know why people don't call them out on this more. Their behavior makes perfect sense from a corporate perspective but instead of it being generally recognized that this is a tool for managing competition we're all to read to go down the 'enhanced features / additional security' rabbit hole -- we believe and we waste lots of productive energy on this. However, the rest of the world has a job to do, and its not being buried in a proprietary ecosystem. So just use the NTFS drivers as read-only and leave it at that (because, let's face it, if we were to use it for write then there's no guarantee that we'd suddenly find subtle data errors creeping in the next time a system update happens).

[MacroRodent]

NTFS-3G

Sounds like the author had not heard of NTFS-3G, which has worked pretty problem-free for years. Mounting, reading and writing NTFS works just fine, and it comes with all major distros. OK, it is a FUSE-based system, but that is transparent to users, and the performance is quite OK, unless you for some reason want to use it as your main FS (and why would one want to do that on a Linux system?).

[Bill 21]

Re: NTFS-3G

Regrettably, the answer's "because"

[Norman Nescio]

Re: NTFS-3G

Sounds like the author had not heard of NTFS-3G, which has worked pretty problem-free* for years. Mounting, reading and writing NTFS works just fine**, and it comes with all major distros. OK, it is a FUSE-based system, but that is transparent to users, and the performance is quite OK**, unless you for some reason want to use it as your main FS (and why would one want to do that on a Linux system?).

*For values of 'problem-free' less than 100%

**For values of 'just fine' less than 100%

***For values of OK less than 100%

NTFS-3G is by no means problem-free. There are, for example, NTFS on-disk structures it cannot cope with - for examples, see the manual regarding other types of reparse points. (It's also worth reading the NTFS-3G FAQ for gotchas) Effectively, NTFS is still a moving target.

For many people, NTFS-3G is 'just good enough', if, for example, you are rescuing some files, or transferring a limited number from one system to another; but as the open-source writers are not the maintainers of the original, they can always be caught out,

As for performance, it's FUSE. I did an backup rsync by mistake from a Linux system to an external NTFS drive. Slow as molasses. Certainly not good at creating the myriads of files. Having stopped it and used ext4 instead, things were far faster. This is not to say NTFS-3G as the target filesystem was unusable, just considerably slower.

I am very glad NTFS-3G exists. It has allowed me to help out friends in need many times, and I am very grateful to the stalwart developers and maintainers: theirs is an unenviable and unending task. For certain use-cases, if you live within its limitations, it is great, and certainly a wonderful addition to my toolbox. However, it is not an officially blessed driver from Microsoft, with all the consequences that entails.

[MacroRodent]

Re: NTFS-3G

I have used it largely for handling NTFS-formatted removable media, which i often find has to be readable on Windows as well. Also for accessing a Windows partition years ago when I dual-booted (no longer doing so, as VM:s are a more flexible solution, so I now give the whole machine to Linux). Never problems with these tasks, but I agree more extensive usage might expose problems.

Not blessed by MS is not so relevant. The popular Samba file server software isn't either, it was created by a combination of reverse-engineering and reading incomplete MS documentation. MS is not in the business of helping open source interoperate, I don't expect them to ever help any NTFS project, at least not on terms that would be acceptable to the Linux kernel guys.

[Lunatic Looking For Asylum]

Beware Obvious Click Bait articles - I should know better

Who cares really. I've been using Linux for almost three decades and can honestly say I've never wanted to write to an NTFS volume. I think NTFS write is an edge case. The existing read only NTFS driver is more than adequate, if you want to write something to an NTFS volume then go and buy a Windows licence. If a project says it has to use NTFS I doubt very much it will say it also has to use Linux. Right tool for the job etc.

I'm struggling to think of a case where you would need NTFS write with Linux as well or even where there is a need for write to co-exist. Just read the data and store it on ext4 etc. or as mentioned in a previous post use SMBFS or NFS. NTFS really is a non issue. If NTFS is the solution, you have been asked the wrong question.

We then come to the politics. M$ will _NEVER_ release NTFS source code. They could release a kernel module or blob so that anybody that wanted write functionality could have it but why would they bother, almost nobody would use it so it's a lot of effort for nothing really.

Releasing the source code isn't as simple as dropping a tarchive out and it's done. There will be a lot of work needed to modify the low level calls to the drivers, problems with endianness, sector translations etc. They would also have to visit every line of code and check that everything is perfectly formatted, logically correct and well written and they would probably have to strip the comments out as well then make sure that it all compiles cleanly on GCC and LLVM and performs in a native Linux environment. They will then release it and some poor OS developer will have it land on their lap and try and make sense of it, make sure that there are no hidden nasties in the code (intentional or unintentional) and then test it thoroughly before letting it free to the community.

Once it's out there, you'll get the sad people poring over the code and criticising every line of it - "oooh look they've used camelCase on line 454 here and CamelCase on line 427 - that's BAAAD...." and even if it's perfectly written, as soon as one bug is found the whole thing will be spread all over the net as M$ flusterCluck and nobody will use it again ever...

I'm still struggling to think why we need it and what the positives are for M$ - as I said click bait article.

[Binraider]

2 weeks ago in response to the paragon maintainer story, I said more or less the same thing as the story.

MS own the specs and sources; they are by far best placed to release the Info needed to allow a rewrite without having to reverse engineer behaviour.

I suspect a major reason they haven’t is that the spec is old; and releasing the full detail will show up vulnerabilities. there’s every possibility the authors are long since moved on too; and as such they may have be maintained by inheritors rather than by designers.

But yes, I’d be vaguely impressed if they were made available. It doesn’t represent a commercial edge to hide these particular sources so there seems to be little reason not to.

[daleto]

There are different ACL

Hi

POSIX has its way to control ACL using ugo (usdr/group/other)

NTFS uses a much more advance ACL

Adopting NTFS to Linux might affecting the way how to use ACL

I am an old linux/Unix person, but admit that Windows ACL is much more flexibel

The next things comes with how I/O are done insider the kernel

XFS has failed to support bigger block size for the filsystem. I guess because the kernel operates in 4K segments

In other words an I/O that are lager than 4K will be choped up to 4K memory segments, placed in a queue to the elevator which tries to glue it together again

I used to compare JFS on AIX where an application used 256K write requests, but the JFS on AIX write to the hard disk was at maximum 80K, to this none fragmented file?

If I instead mount it using SANergy which is similar to newer version of NFS where one can split up meta and data trafic, going to same type of filesystem (JFS on AIX), I got 256K on each write request.

The only explanation I have for this is that all POSIX are most likely performing the same cut-and-paste for I/O wich in my opinion is waste of CPU resources.

Why explaining this here?

Cause when I compared filesystem performance between NTFS and at that time available filesystems, NTFS was far way faster, which I guess is because of that they might not use this cut-and-paste in the kernel.

IF Linux NTFS supports bigger blocks for the filsystem, than I guess this will happen here too?

XFS says (if I remember it correctly) that if one recompile the Linux kernel to use bigger than 4K, than we can support this as well.

But 4K is used by memory allocations etc and that affects may affects many other things too

Porting NTFS that has the same charasteristics isn't as easy as it sounds

Different I/O mechanism, different ACL

Thanks for this article

[daleto]

Fuse added fix for better write performance

NTFS-3g so far is proven to better then the new in-kernel ntfs driver in terms of stability and being more bug-free.

Micro kernel design might not be so bad?

[Peter Gathercole]

I'm confused

I know, easily done nowadays (confusing me, I mean).

GNU/Linux has absolutely got NTFS support that allows read and write access as modules that are added to the kernel after boot, and also as a Fuse filesystem.

What it lacked before is NTFS read/write support as in the kernel itself. That is what the Paragon software adds to the kernel. And I'm not sure that we actually need it.

I can consider some situations where it may be useful, possibly in embedded devices (although it's probably a bit big for that). What it would mainly provide would be to have NTFS filesystems that were mounted to a system before it was fully running, as you might expect if you were booting a Linux system from an NTFS filesystem. But how many people really want to do that?