Hacker News new | past | comments | ask | show | jobs | submit login
exFAT in the Linux kernel (microsoft.com)
454 points by UkiahSmith on Aug 28, 2019 | hide | past | favorite | 271 comments



I'll believe Microsoft loves Linux when I can install Microsoft Office in my desktop Linux machine. Everything else is marketing.

Edit: to expand, the corporate world runs on Excel/PowerPoint/Outlook. Microsoft milks them on Windows/Office licensing and is very aggressive against organizations that try to do Linux deployments. They're a convicted monopolist that got away easily when at some point the option on the table was to split OS and Office into two different companies. If they want to continue to use Office to keep Windows dominant that's their strategy. WSL and their other "love Linux" efforts are all about making sure developers stay on Windows. If you want to do all that fine but don't patronize us by then claiming you love Linux.


I would recommend against using Office 365 web ui for word/excel etc, I'd also recommend avoiding the MacOS Office build for the same reason:

They're both buggy as hell, the type of bugs that will make your document render in unintended ways when somebody opens it on the other side. At some point, Word for Mac decided to remove whitespace between words on my resume - I couldn't see them and generally exported to PDF, but I didn't hear back from prospective jobs that asked for a word format specifically.

Office 365 Web and desktop application really need a complete revamp, they have reproducible bugs and horrible UI/UX in edge cases.

I also really hate when Microsoft decides I want to store my sensitive data on their cloud for no apparent reason despite saving to local disk, it really seems like a 'whoops we accidentally did this but you should try it!' kind of move from MS. This is the perfect example of a monolithic application with chronic feature creep.


> I didn't hear back from prospective jobs that asked for a word format specifically

This almost always means you're speaking to a recruiter who is going to strip your identity from the resume so they can hold it hostage for an agreed commission. This also puts your submission at a disadvantage because you're automatically 10% more expensive than the next guy.

If you're submitting your job application, double check that you're submitting through the company's preferred channel. Look for the job on their own website. And be suspicious when they request an editable file format.


Thanks for the tip, I learnt this a long time ago, I've also moved into an industry where dealing with recruiters would probably indicate other issues like lack of ability to research prospective employers, etc.

I'm glad it's becoming more common knowledge though. Recruiters have long been redundant compared to job search websites like seek.


In some cases yes, but the majority of jobs are never formally advertised.

Where do you think MD (Managing Directors), CxOs and other high level people go for jobs?

The majority of them will get their next position through who they know, not a job board.


A good one though, very much worth it.


I was lucky enough that I never had to apply for a job with such requirements, but how does one learn about stuff like this?


In my case, it was what I discovered after having it explained to me during a job interview why the salary I was asking for wasn’t feasible. The company straight up said they couldn’t afford on top of the recruiter’s commission. I had assumed the recruiter was representing the company, but after that experience I spent time learning about the murky side of recruitment.


I had a really awesome boss who probably overshared about stuff like this but a lot of it also comes with general distrust in the business world. People are out to make a buck and they'll do it anyway they can. Every time you come across a business, consider how it works and where their profit comes from.

A tactic I've used in the past is buying a burner number (prepaid sim), called recruiters with a fake name, number and resume and asked them to provide details about the job which many of them name completely. The ones that don't generally indicate that other recruiters do exactly what I'm doing to them in order to steal clients.

I don't feel bad about screwing over an industry which has no place in the modern world, particularly when they're opportunistically trying to make a buck from me and/or my future company while adding very, very little value :)


Have friends that work/have worked as a recruiter :)

PDF files aren't safe either. Even if you gave them a JPEG, they'd just transcribe your data into their own little CV template.


Sure they can, but they'll still try for a DOC file because that makes their lives easier.

The answer isn't to try to "beat" the recruiter with PDFs and JPEGs; the answer is to run away and find a more direct path to the employer.


> I'd also recommend avoiding the MacOS Office build for the same reason

I never understood why they made it a effectively a different product. (Once you go past the trivial interaction) Different features, different behaviours, different problems. Why is excel for Mac not the same engine with a different UI?


Probably likely to do with the toolkit used to build the Windows version and inability or unwillingness to port it to Mac.

They really need to start from scratch and build solid, easily testable product because the current methodology doesn't work.

I especially love the bugs where when in one specific track changes mode typing in the comments section drops keys, or when using 'read aloud' the voice randomly changes gender. Office 365 on a Windows 10 LTSC virtualised host w/ no other software.


Honestly, Libre Office has better compatibility between office versions, then MS Office.

I also like Abi Word, but it's only a word processor, not a full suite. I keep it around because it will open damn near any word processing format.


> They really need to start from scratch and build solid, easily testable product because the current methodology doesn't work.

Isn't that exactly what you're trying to do with the web version


I actually thought that this is what they were doing:

https://www.zdnet.com/article/how-microsoft-is-taking-on-the...


Hmm... Maybe I'm using something from before this project was finished. The support pages claim that the latest versions have the same behaviours between Win and Mac for the features I was interested in. May be time to upgrade.


Given that Microsoft is a for-profit company, what's the financial justification for writing office for Linux on desktop? The market share isn't even a rounding error and you're talking about them investing tens of millions minimum into the development.

The fact they have office on Android, iOS, and osx proves it's about market share and common sense, not some fabricated hatred or fear of Linux on the desktop.


> what's the financial justification for writing office for Linux on desktop?

Well, the same argument could be applied to their open source contributions. Considering that [canonical corporate view is that] patented ideas are expensive to produce, and that exFAT is patented, what's the financial justication for open sourcing it (the specs or else)?

The argument is ultimately against "Windows love Linux". The thesis of the parent is just that they're doing unsubstantial moves (or worse, manipulative ones), and that if they "truly loved Linux", they would do something substantial, which, for the corporate world, is to support Office on Linux.


This argument can very easily be used against people who claim they love Linux and FOSS but they never really seem to be willing to part with the same amount of money other people are willing to give MS. Donating $300 for every copy of Libre Office you use would go a long way to making MS Office on Linux completely unneeded. Same goes for donating similar amounts for every Linux you install. I guess everyone's love has limits and it's usually money.

As for MS's contribution to open source you have to consider the difference in magnitude between the revenue that Office generates and the revenue that their other open source contributions do (would?). Then consider also the development effort involved.

It looks like OP just wants exactly what MS doesn't offer completely disregarding the alternatives, and I'm pretty sure they would not be willing to pay the markup for the extra development involved.


They don't have a mobile OS anymore. Apple is hardware-limited; there's a much larger buy-in required if you wanted to switch from Windows to MacOS than to Linux. And companies could save a LOT of money on computers if all they needed for software was Office and that were available on Linux.


yeah but the money they'd save is money that would go to microsoft through windows licenses and support, why would they want to lose that?


> you're talking about them > investing tens of millions > minimum into the development

Do you have a source for this hyperbole??


That’s what, a few dozen devs and infrastructure, probably? Not at all outlandish.


This seems oddly specific and arbitrary. What if the engineering effort in porting it over is not economically viable?


Given the direction they're going with the web versions of Office 365, it's entirely possible that the future involves them deciding porting stuff to the Windows version is not economically viable and relegate the native apps to legacy support like IE.


I've been 100% Office 365 web apps on Linux Chromium since around February. So far, I haven't hit a situation that has required spinning up my Windows 10 VM.


What do you do when you want to edit a document or compose an email offline?


I don't know. I'm in a US coastal megalopolis covered in cell data towers, so I haven't encountered that issue yet. I'm sure it will happen at some point, but I think the web apps have offline capability.


I am guessing Office 365 has some sort of offline capabilities. Maybe not?

For the email, can’t you write it anywhere? Difference is it’s not officially a draft in an email client.


Office 365 allows you to install the latest version of word excel ... too in addition to their web versions. It’s actually the superior part of the offering, even though Hm tends to focus on the web part (why superior ? Automatic upgrade to latest so no mismatch of versions inside your company, no expensive acquisition cost merely add less than ten euro a month to add a new user, disable the next month if it didn't work out).

The installer fails on wine last time I checked, but porting an installation from windows works fine, and the account and license connection is fine too.


Office 365 is 700 things all with the same name.

Office 365 web apps (free), Office 365 the monthly updating desktop Windows version of MSOffice (not free)


Outlook webmail works fine offline. You need to enable some setting though.


That's actually kind of happened. One note program is no longer part of Microsoft office anymore. Instead you are suppose to use the onenote app. It's a hassle but you can install the old program and it's still much better than the app


It's more likely that Microsoft Office will become 100% "cloud" based software.


That's never going to happen. The US government is a huge Microsoft Office customer, and national security-sensitive documents cannot be placed in a cloud that's not completely controlled by the government.


Sounds like they'd build a Gov-Office365 cloud if there's a deal just like AWS Govcloud.



There’s already a US Government Community Cloud for moderate risk (.com) and a US GovCloid for higher risk (.us) workloads.

As with all things multicloud and Microsoft, it’s easy to get into trouble in the community cloud.


Have you heard of AWG Gov tier?


Maybe they don't want to deal with how shitty a graphical desktop environment is on Linux. I've largely given up on Linux with the exception of the terminal because of this. There are just too many graphical quirks and glitches.

I know that Linux is merely a kernel but I wonder how things would have played out if the kernel contained some form of graphical environment. There are so many competing environments that you have massive choice but also nothing that really works well either.


Pretty sure porting SQL Server to Linux was more than marketing.


Run SQL server on more cloud providers that way


Okay but this is the future of our industry. Are they just supposed to roll over or die?

Shouldn’t we appreciate that after decades of bad blood they’ve decided to now wise up? Shouldn’t we give the benefit of the doubt?

Companies have to face it. Community-which was once considered auxiliary bullshit that needed lip service—has now become the de-facto requirement for how to engage users and customers in a technology business. Now that we see them doing it we want to criticize them?


Companies are not people. They are not owed any second chances, olive branches, benefits of doubt.

Nothing is stopping any of MSFT's OSS-Forward ideas from being brought up by capable people in other contexts.

> Community-which was once considered auxiliary bullshit that needed lip service—has now become the de-facto requirement for how to engage users and customers in a technology business.

I don't know what you're referring to; I don't see any difference in community engagement in the industry on the whole compared to 10 years ago.


The only move towards "community" from big companies I've seen so far is that nowadays you can make a ruckus on twitter to make them care.


> Shouldn’t we appreciate that after decades of bad blood they’ve decided to now wise up? Shouldn’t we give the benefit of the doubt?

If they didn't supported SQL server o Linux then no one would use it, as using MS Windows on VMs forces an additional premium on deployment costs.

As I see it, supporting Linux on a specific use case is not a sign of wising up, but a desperate concession to try to stay relevant.


> As I see it, supporting Linux on a specific use case is not a sign of wising up, but a desperate concession to try to stay relevant.

Those are really the same thing, "wising up" and "trying to stay relevant".


Of course. Similarly, Linux is run more than Windows on Azure.


This also means the majority of OSes operating on a Microsoft proprietary cloud arch are based around a kernel whose governance is becoming increasingly influenced by - you guessed it - Microsoft.


I would settle for just Skype for Business on Linux. No excuses why it shouldn't be on there... and yet it isn't.


Skype for Business has been replaced by teams which is not much better and pretty much still an awful chat application.


Teams is worse in that it appears to actually be Electron instead of the semi-native (I don't think it's pure Win32, judging by the UI latency and appearance) Skype for Business.


Being awful is Skype's killer feature. You can't carpet-bomb with pings, you can't comfortably copy and paste code or logs into it, half of devs who use linux don't bother running it on a VM and instead use flaky third-party clients, it eats messages randomly, so long story short - people generally avoid it except for trivial shit. And on the odd day when they don't, I get a copy on Outlook anyway.

I've read some horror stories about Slack, and I shudder to think what would happen if my company moved to it.


only for the cloud version.

S4B server edition is still in active dev


You could use office 365 on your linux machine.


Offering a cloud service to a linux user is like tempting a vegetarian with meat.

Linux users often choose Linux because they want to be in control of their data and devices, cloud services are about giving control of that away to companies.


Linux is a kernel, it’s not a way of life. Chromebook and Android users are Linux users. People who use web applications that run on Linux servers are Linux users. Developers who use Vagrant or Docker on Mac OS, or Windows Subsystem for Linux on Windows, are Linux users.

Microsoft may or may not “love” the tiny subset of users who run Linux-based desktops and are unwilling to use web applications. But that doesn’t mean they don’t “love” Linux as a foundational technology of modern computing.


You're being pedantic. "Linux users" in a post like this clearly means a specific thing. Context matters.


It means a specific thing, but the specific thing that it means is insignificant. The basic argument that “if Microsoft really cared about Linux they’d write native Office apps for Linux desktops” is myopic and, in itself, ignores the broader context of what makes Linux relevant in the first place. As you say, context matters—and context is exactly what I was providing.


Agree with you 100%. On a related note, it strikes me as wrong that so much of the discourse about Linux revolves around its suitability for general consumer desktop use.

People always talk about “the year of the Linux desktop” (and not always ironically!) but honestly, whether a random person can play 4k Netflix on their Linux laptop is like 0.1% of why Linux matters in the world.


> Linux is a kernel, it’s not a way of life.

You're being disingenuous. Linux users are users of Linux and software that runs on Linux.

In the context of service deployment, this clearly refers to users who need to deploy Linux VMs to run software that runs on Linux.

The kernel is irrelevant. This use case clearly refers to Linux distributions, which nowadays probably means Debian or a Debian-based distro. So obviously in this context a Linux users means someone who needs to deploy a Debian-based distro on VMs to deploy and run software on it. In fact, more often than not when people refer to running Linux they are actually implying that they need to deploy and run software on a specific ecosystem, such as a combination of a package manager, package repository, service manager, and even f Directory tree layout.

This is not a way of life or a cult. It's a technical requirement that must be met for people to do their job.


> Linux users are users of Linux and software that runs on Linux.

That's exactly the point I was making. People who run Debian-based desktops are a tiny subset of that group. Hell, I have a Linux desktop and not even I run Debian.


>Linux users often choose Linux because they want to be in control of their data and devices, cloud services are about giving control of that away to companies.

You may have been able to get away with that kind of broad generalization a decade ago (actually mo -- that's not true, AWS and Canonical's push to cloud started MORE than a decade ago. Google too. Let's call it 15 years ago), but that is certainly not the case now.

And frankly, I think putting Linux users in a box where only FOSS drivers are allowed (even though Nvidia GPUs are doing much of the best work for CUDA) and to use it you have to be lock-step with an ideology that very few people will ever be able to be "pure" enough to follow completely (assuming they want to follow that ideology in the first place), does the Linux ecosystem a major disservice.

I would posit that the vast majority of developers and end-users who access and use Linux on a daily basis are doing it at least in part through a cloud service -- whether its a cheap VPS or a large cluster of machines.

*Disclosure: I work at Microsoft on Azure. These are my thoughts and do not represent those of others. I primarily use a Mac but have been playing with/arguing about Linux for 20 years -- going back to when I was 15 years old. I also use WSL on Windows.


I also work on Azure, hi!

I have mixed feelings about closed-source blobs tainting my kernel. AMD did a terrific job open-sourcing their driver stack. I use Nvidia for deep learning out of necessity but I resent it, as I'm completely locked into X11 because they can't be assed to implement GBM. Nvidia has also gotten a stranglehold over the DL and HPC markets with CUDA, causing a chicken-and-egg problem where OpenCL isn't well supported because none of the cards that require it have serious power, and it's not worth AMD putting lots of money into the HPC market because it's locked into CUDA.

NVidia needs Linux more than Linux needs NVidia (imagine competing on HPC while running Windows-only!) And yet, they contribute nothing to the kernel and stymie efforts towards unified standards like GBM.

it's shades of DirectX all over again. that's the problem with tainting.


Well we are talking about using Linux on a desktop machine exclusively, only then a demand for a non-cloud version of MS Office would make sense, I think. This subset of people are probably more likely to care about control of their data that just some people that dual-boot or use linux on their android phone.


I think a significant number of Linux users don't care about this.


That depends. If they are a Linux user because Linux runs on their server, router, mobile phone or what not. Then you are probably right.

People that go through the trouble of installing Linux on a Desktop Machine, where Windows or MacOS was probably pre-installed, will either dual boot or use Linux exclusively. If they dual-boot, they will probably just use Office on the non-Linux System and not care.

But if they use Linux exclusively they will very probably care about the freedom and control Linux brings.


I still disagree. I used Linux almost exclusively because it made my job easier and I don't like switching between operating systems, and I use cloud services like Gmail and Dropbox. I also use the free version of onenote from ms. I just use whatever is convenient.

Then I switched jobs, they gave me a new MacBook that doesn't run Linux so I use osx.

Ten years ago I cared about privacy, now I feel like that ship has sailed. My own privacy is not that important. The privacy of my fellow citizens is gone. I mourn it kind of, but you can't close Pandora's box.


There are a gigantic number of people whose university or corporate IT department installs Linux for them on standard issue hardware which they use every day for their job. E.g. big companies whose programmers all develop on Linux machines, or most academics in programming-adjacent fields like statistics or machine learning.

Another category is people who installed it themselves because they like it better, but for non-ideological reasons. Like, a large chunk of programmers feel much more comfortable on Linux than Windows.

(I’m not sure what to make of your point about dual booting — it’s hard to believe anyone would “not care” about having to do a full reboot every time they wanted to do something as common as view an office document. Now “let me finish reading that interview candidate’s resume” goes from something I can alt-tab to while compiling into an ordeal of saving all my open work, rebooting, downloading and opening and reading it, rebooting again, reopening all my work...)


I fall into the another category. I use Linux because I like how I can customise heavily how it looks and have control over when updates are installed. I do some programming as well which does admittedly work better for me on Linux. I don't really care about the ideology. I use a Nvidia graphics card with closed source drivers because I want the best performance. Free and open source is nice, because I like how that works as an ecosystem but I'm never going to turn something I want down because its closed source. I considered dual booting for gaming but honestly enough works and as you say about rebooting just to open an office document, I don't feel like doing that just to play one game and then go back.


> want to be in control of their data and devices,

The other day there was a thread on HN about which laptop is great for Linux. Tons of people recommended Lenovo, a guy from Redhat said it's even the brand that's used internally and I got a little depressed and smh. Nobody mentioned superfish. Those glorious days of Linux are gone.


Redhat internally uses Thinkpads.

Superfish was on Ideapads, and it was bundled Windows software.

While both are by Lenovo, they are separate product lines.


"We are a totally trustworthy company, we screw only some of our customers."


In that segment? I would wonder something is wrong, if they didn't.

Ideapads are in market segment, that is extremely price sensitive, but not quality-sensitive. If the competition would cost 10 bucks less, it would be a huge win for them, so everyone tries to minimize price while preserving margins as much as possible.

In the end, these devices do ship with Windows, which has privacy problems anyway.

All this is moot, when you put Linux on it, which was the context we were talking about.


> In the end, these devices do ship with Windows, which has privacy problems anyway.

It does, but those are "theoretical" in comparison to the degree Lenovo stooped on, it doesn't excuse Superfish at all. Windows isn't a free for all platform, yet.

> All this is moot, when you put Linux on it, which was the context we were talking about.

_That_ time. I'll copy a response I provided two months ago as Lenovo on HN has become a seriously annoying groundhog day for me:

I can't know that whatever harmful and irrational environment that led to Superfish in IdeaPad won't affect ThinkPads in the future. Even in the most generous understanding where IdeaPad is a different, physically separate branch of the company, and Superfish was an act of incompetence and not outright malice I can't be expected to keep up with the insider intrigue of the company to notice any changes that could negatively affect me. More importantly, leadership is still responsible for setting irrational environment that lead to Superfish, whatever that environment was. This is a multi-billion dollar company, there is no excuse for such incompetence.


Every single company did stupid things in the past (not just CE/ITC companies; but the builder that built your house too, for example). You can be outraged by many things all of them did. By choosing, what you are or are not outraged, you are just rationalizing your preferences.

Was superfish stupid mistake? Yes, it was. Does it mean you should condemn the company for the rest of eternity? Probably not.


> Every single company did stupid things in the past

Perhaps there is some boundary that shouldn't be crossed. You seem to be arguing there isn't.

Perhaps global and successful companies should exercise some due diligence and have a department that would control quality and firewall wacky ideas.

Perhaps we as CEs should hold companies responsible so that they stop perpetuating never ending shenanigans that this industry is known for. If even we don't boycott bad actors how can we expect normal users to do it.

> Does it mean you should condemn the company for the rest of eternity? Probably not.

If they showed some contrition perhaps? Name and fire entire chain of people responsible for it and donate a year of their net income to foreign FOSS organizations - in ideal world the company should've gone bankrupt, so the penance should be hard. Then I'd be more ready to believe them they would do better in the future.


To each his own, in my opinion consumer market for laptops being a trash fire doesn't excuse putting malware on them.


Yes, it was stupid. Should that mean you are never going to buy thinkpad again? Probably not.


The whole point of being vegetarian is to not eat meat. That in no way is analogous to Linux and cloud stuff.

You seem to be talking about a subset of Linux users. And then a subset of that subset who find cloud apps as bad as vegetarians may find meat.


OT: I only need Outlook or at least a decent Groupware Client for EWS (EWS for Outlook Online Sync/the customer version)/office365. That does not have sync issues, etc. Bonus: if it can sync Contacts with Linphone. And I would pay for it after a 10days testing period. electron version is ok, but not preferred.


> I'll believe Microsoft loves Linux when I can install Microsoft Office in my desktop Linux machine.

I'll believe Microsoft loves Linux when I can install Windows after Linux and it doesn't mess up the bootloader as if it were the only game in town.


I'm not sure that "desktop Linux machine" is really a thing. I hate to feel that Linux has won servers and IoT, maybe mobile, but clearly lost desktop. Desktop is a dying thing by itself, so not bad after all.


they ported SQL Server but I doubt they would ever port Office. It would make more sense to use your browser to access online Office.


The IMHO extremely relevant point is in the very end of the actual specifications:

https://docs.microsoft.com/it-it/windows/win32/fileio/exfat-...

>26-Aug-2019 Seventh release of the Basic Specification, which includes the following changes:

Updated legal terms pertaining to the specification, including:

Removal of Microsoft Confidential notice

Removal of Microsoft Corporation Technical Documentation License Agreement section

Updated copyright notice to 2019

Till now the documentation wasn't AFAIK publicly available or it was anyway "restricted".


Skimming through the doc, it looks like a modification of FAT32 with some (noticeably more complex) extensions, and not anything resembling what you might envision a "FAT64" to be; a FAT is still a linked list of 32-bit entries, there can still be two of them, and directory entries are still 32 bytes each (although defined in a rather complex way.) There's a separate allocation bitmap, which seems redundant and a waste of space (and code) given that the FAT itself already contains the information in previous versions.

I've written a FAT12/16/32 driver for an embedded system before; and oddly enough, the documentation I had at the time was far easier to understand than this one. I didn't expect exFAT to be this complex, or perhaps the doc is just excessively verbose --- for example, the whole first paragraph of the 4th section is saying nothing more than "there may be 1 or 2 FATs, one after the other, and NumberOfFats is the field that says how many there are" except it takes 5 sentences to do so.

That said, this spec is written in a style that's easier to understand than a lot of the other "open specifications" docs that MS has released, but I suspect that's because this one was once an actual NDA'd implementer's document rather than the others which I've heard may have been released purely for legal reasons related to the antitrust than anything else.

Now if only they'll do this for NTFS...


With how slow modern day NTFS is for things like node_modules (lots of nested directories), I can't help but hope some better file system will become widespread on the next major Windows release.


Most of the performance issues are not NTFS so much as the prevalence of file-system filter drivers. This is why WSL had so many file-system performance issues despite the fact that it was using a custom FS. IIRC this also motivated the move to WSL2 from what I read, since a virtual machine can use a small set of files to store a whole FS.

EDIT: See here - https://github.com/Microsoft/WSL/issues/873#issuecomment-425...


The big deal here is the patent grant. FAT related patents were held over linux like a gun in the Ballmer years. Its a reason Linux was stuck with 8.3 length filenames for so long.


Agreed! I made reference to the patent in the exFAT Wikipedia page for years ago and it was reverted within hours.

When Apple was at death's door (around the arrival of OS X) they were very open about protocols and formats (e.g. JPG, mp3, ssh, whatever) and happily paid the Danegeld so their customers could encode mp3s and RTF files etc. They proclaimed how important open formats were and managed somehow to get unprotected files in their music store. I thought to myself that they would lose "religion" if they ever achieved any success and indeed, they don't really care so much. They don't even oppose it, just don't care.

So I wonder if this is MS seeing that exFat might be replaced just to save a few pennies of BOM cost and that they need to be in the thick of IoT things, having lost the embedded OS competition. And while exFat is hardly a big money earner, staying relevant matters. With linux beating Ones in the Azure cloud, every bit helps.

BTW I'm a long term apple user; my comments about the waxing and waning of their commitment to openness is a dispassionate observation.


While an interesting thought I believe this isn’t it because whatever new fs would need to work on windows and be standardized.


They fired the gun. Linux had to rework the FAT32 implementation in the kernel when TomTom got sued for not paying the Microsoft tax and lost.


TomTom didn't lose; they capitulated.


>> Linux was stuck with 8.3 length filenames for so long.

How so? I never had that problem.


.... on FAT partitions


Sorry, dumb question. :)

I guess I never noticed a problem because either I mostly use ext3/4, or I am so used to using legally dubious drivers that I forgot they were there.

I mean, who ever gives you a FAT formatted disk? I don't have floppy drive anyway :p


Cameras. High capacity removable flash media used by digital still and video cameras is often formatted as exFAT. Despite MS' licensing, FAT/FAT32/exFAT are essentially the "file systems of interchange" for removable flash media.

More info at: https://fossbytes.com/fat32-vs-ntfs-vs-exfat-difference-thre...


This was also a huge problem for Android. At some point Android started using MTP instead of appearing as a mass storage device, which made for several technical wins. Most importantly, internal memory could be one contiguous block formatted as ext3. But this happened well before MTP was properly supported in Windows, which predictably made things difficult for a lot of users. (This was back when Android had competition, so that was important).

If I was a betting type, I'd wager that Microsoft's FAT patents (and various legal settlements where they got to say "Linux infringes on hundreds of patents" without ever needing to specify the patents) had a lot to do with that change, since, with the exception of a few manufacturers who continued to ship devices with external SD slots, and USB-OTG which is kind of a bonus feature, Android finally didn't need FAT for interoperability.


The SDXC standard (SD cards > 32GiB) specifies exFAT as the standardised Filesystem.


That's because no one has bothered to make a better FS for that sort of thing. Open FS's all insist on having UNIX's archaic permission system, which is useless at best and at worst actually harmful to moving files between systems, and they aren't as absurdly simple as FAT.


Linux itself creates a FAT partition for /boot, because the UEFI standard supports loading from FAT.

As well, if you ever look at a live-CD (like the installers for most modern Linux distros are), it will usually be using SYSLINUX, which limits itself to 8.3 filenames, for various good reasons to do with floppies, PXE, and ISO9660. Mount an Ubuntu CD in Windows and you'll see them.


> UEFI standard supports > loading from FAT.

Actually UEFI REQUIRES the bootable partition to be FAT (not not exFAT).

UEFI also requires the executables to be in the COFF format ( instead of the normal ELF one ) too.


What makes ELF normal? Funnily COFF also comes from Unix.


And IIRC, FAT being in the UEFI specification kind of solved the patent issue over FAT being in the Linux kernel tree.


EFI System and /boot are often the same partition, but they don't have to be, GRUB can boot a kernel and initramfs off most common linux filesystems (and less common ones like xfs, hfs and reiser), it can even decrypt a LUKS partition.


A liveusb using FAT+SYSLINUX/Grub4DOS/whatever has another really good reason: it's readable (and probably writable) from anywhere.

I.e. my EDC flash disk has FAT32-formatted first partition, yet is bootable. This is surprisingly versatile.


If you ever interact with people running Windows (or even OSX) all of their thumbdrives will be formatted with FAT unless they recently used it make a Linux boot disk after installing a later version of windows 10 that has extx support and didn't notice.

If OSX and EFI had out of the box support for Linux filesystems we could at least stop using FAT on pendrives. I personally will look forward to that.


Later versions of Win10 have ext* support?


No. He might be confusing it with WSL2 which uses VHD files with ext4 file systems hosted in them, but those are only accessed by a Linux kernel under virtualization, never by Windows itself.


Most flash drives and SD cards are still FAT formatted by default.


SDCards, USB sticks...


USB flash disks are almost always exfat formatted.


[citation-needed]

Most USB flash disks I've seen (and I've seen many) are even nowadays mostly FAT32. "Portable drives" aka USB-to-SATA are shipped preformatted to NTFS, and only with the advent of SDXC did preformatted exFAT start to appear.

(It has to do with driver support, I would guess - exFAT works in Win7+, but as long as the world runs on WinXP kiosks (ugh), vendors go with the lowest common denominator. Note that a flash disk will go places - what's your car entertainment system running? Linux? Android? Windows CE? Anfient Embedded Monftrofity? Will it support exFAT? Unlikely.)


Aren't the FAT patents about to expire anyway?


Some of the earliest ones, surely, but presumably Microsoft also continued to patent later innovations: all of the "extended" in ExFAT.


This seems to be one of the patents involved: https://worldwide.espacenet.com/publicationDetails/biblio?CC...

I'm not sure what's the expiry on 2008 submission.


20 years from invention? Which looks like ~2004 skimming the 2008 filing.


The original ones from the MSDOS days should have long expired. The one with VFAT/FAT32 from the Windows 95 era should have expired a few years back. I think exFAT is probably the last one remaining.


Yes I forgot that exFAT is a whole new thing from 2008.


Huh? I'm sure I've yes long file names on vfat formatted disks for years...


does this mean we can finally get MSC back and not be stuck with the giant turd that is MTP?


I don't mean for this question to sound inflammatory but I can't help but think: why does this matter? You can already use exFAT on Linux via FUSE, and as pointed out by VentureBeat:

> To be clear, Microsoft isn’t open-sourcing exFAT — but it is making sure anyone building with Linux can use it. 'It’s important to us that the Linux community can make use of exFAT included in the Linux kernel with confidence.'

exFAT is still not open source, and still can't be distributed with the kernel. So what changed? It will now be easier to integrate it into a custom kernel instead of using FUSE? Why is this better than the current situation, where you'd install exfat-utils for example? Are there benefits of using a kernel module that I'm missing which makes this important news?


It will be in the "real" kernel soon, here is a patch I submitted adding it: https://lore.kernel.org/lkml/20190828160817.6250-1-gregkh@li...

And yes, this is _much_ better than using a FUSE interface to the filesystem.


For the 99% of people reading who don't know the kernel as well as you (seriously) can you explain why this is the case?

Does a userspace filesystem driver lose a lot of performance to context switching, or is there something unique to filesystems that slows them down in user space?


> Does a userspace filesystem loose a lot of performance...

Yes.

Try the two versions out, and see for yourself if you are curious.


I use a FUSE filesystem on linux- not ExFAT, but as an S3 interface. I often see 1+Gigabyte/sec throughput over the wire from simple multithreaded IO operations. FUSE can be very fast.


For one thing, that's throughput, not latency.

For another, that's probably a desktop, not a smartphone.


That's correct, my metric for performance is throughput I don't really think a desktop vs a smartphone matters much at this comparison, it's just my observation that when people say a fuse filesystem is slow it's just not well-engineered, not that the fuse API is inherently slow


Run a benchmark that measures more than throughput and you'll see the difference. IOPs are usually poor.


Latency is a nightmare if you perform certain types of tasks on FUSE.


It matters a lot. You can squeeze impressive performance from a smartphone, too, but then you need to dump all that waste heat somewhere, plus you need to take battery capacity into account. In other words, in a desktop, power efficiency is commendable, but not critical.


For copying photos off an SD card you're unlikely to notice any difference on any machine made in the past 10 years, except maybe the fans coming on, but try it on a Raspberry Pi class machine, or doing any kind of random IO and oh boy..

The Spectre/Meltdown situation must have made FUSE much, much worse than it was already. I wonder if anyone did any benchmarks for that


Much faster. Also, FUSE has to be installed by the user, which if it's in the kernel, then it Just Works everywhere. With the spec open and the patent license freely granted, the most common USB flash drive format can now Just Work everywhere.


It also removes the requirement of distros packaging it up and users installing and configuring it. Is it's in the kernel it just works out of the box everywhere.


better because you can be sued for using the FUSE interface?


According to John Gossman "Microsoft Distinguished Engineer & Linux Foundation Board Member" in a related announcement, Microsoft plans to contribute it to the Linux kernel:

We also support the eventual inclusion of a Linux kernel with exFAT support in a future revision of the Open Invention Network’s Linux System Definition, where, once accepted, the code will benefit from the defensive patent commitments of OIN’s 3040+ members and licensees.

Still some patent squirrelly words (they imply you have to sign a patent cross licensing agreement with the OIN - not sure of the implications) and no timeline.

Ref: https://cloudblogs.microsoft.com/opensource/2019/08/28/exfat...

OIN Ref: https://www.openinventionnetwork.com/joining-oin/


Most of the major Linux distributions are directly members or licensees of the OIN already. It looks like any distribution that isn't can get a license by filling out a form, no money needed and royalty-free.

To my reading, the OIN is a "mutually assured destruction" organization for patents around Linux that owns some of the patents outright and basically tries to insure that if any member sues another member or licensee over any of their patents the OIN itself and all the other members should band together and counter-sue over the remainder of the patents. It seems an interesting nuclear option to protect Linux from patent lawsuits.


The list definitely is long but I don't see Debian or Archlinux there: https://www.openinventionnetwork.com/community-of-licensees/


Well at least they did not deny patent issues. As for the driver, a community one is already on the way so they'll probably just piggyback on it.

Or, if they want to contribute an official driver, they should have contribute it at the same time as the spec, IMO.


The big deal here is that they are placing the patent into OIN - I'm probably correct in guessing that it is now the most powerful patent in the OIN arsenal. More electronics/software shops are going to be pressured to join the OIN. Consider sharing files on physical media and, oh, that UEFI basically uses FAT.

Patent trolls are probably collectively experiencing a migraine after this news.


Also hope that uefi will soon use exfat. Still better than plain old fat.


> why does this matter? You can already use exFAT on Linux via FUSE

It means that exFAT will be available everywhere. For instance, right now if you want exFAT support on Fedora or RHEL, you have to enable rpmfusion and install fuse-exfat and exfat-utils from there; once it's in the upstream kernel, it will be available on a default install.


But they can put their own implementation of exfat inside the kernel now without worrying about patents, which is what previously prevented it from being included. I will also help that they will have an official spec to go off of now, so they won’t have to waste time with reverse engineering (plus the legal concerns that comes with that).


FUSE is very slow compared to kernel module.


Citation needed.


I’ve been building SD card based Pi devices and the limiting factor in IO perf is the FUSE exFat implementation. There’s a leaked Samsung internal implementation that is over 2x as fast in my benchmarks. I can’t attest whether it’s the fact it’s FUSE or just other performance optimizations that is the reason though.


The leaked kernel was apparently later open sourced [1], although I can't actually find it on the official samsung website.

[1]: https://www.phoronix.com/scan.php?page=news_item&px=MTQzODQ


I don't think we need a citation for common knowledge about how OS kernels work


Doing things in-kernel is almost always faster. No dealing with complicated context-switching, moving memory back-and-forth, etc. I can't necessarily provide a formal, data-backed citation, but it's pretty well understood.


Because Fuse is slooow - I've had an external Exfat drive mounted on Raspberry Pi 4 and writing large amount of data would pretty much max out the CPU. Ext3/4 has no such issue, I can write to it for days without taxing the CPU. Apparently it's a known problem with Fuse partitions, NFTS-3G has the same problem.


legality and implementation straight on linux kernel would be the first thing I'd say.


exFAT will also be available on other devices like cameras android phones


Care to explain how?

Microsoft isn't giving away it's exFAT patents. They are merely suggesting, that everyone licensed them at Microsoft's own terms. Those terms include becoming a member of Microsoft-sponsored patent ring and agreeing to it's terms and conditions: https://www.openinventionnetwork.com/joining-oin/oin-license.... The conditions can be changed anytime (!!) and the patent license terminates if you sue Microsoft or any other OIN member for any patent violations whatsoever.

Nikon, ZTE, Xiaomi, Samsung and thousands of others (basically all, who were or still are patent-trolled by Microsoft) haven't magically become members of OIN, and it's unclear if they ever will.

Google and Red Hat are members, but neither produces any devices with external SD-cards...


Does this affect anything outside the Linux kernel? Will *BSDs be able to use the specs without patent fears? What about microcontrollers, and the Arduino community?


No. The OIN promise only applies to Linux-based systems.


What defines a "Linux-based system"? Does Android count? Would VMWare's VSphere ESXi hypervisor count? Would a unikernel running on a pared-down Linux rump-kernel (if such a thing were made) count?


OIN works with the Linux kernel community to define a "Linux System Definition" that has tables of packages and functionality covered. New functionality is added every year. It does not include proprietary software or components.

https://www.openinventionnetwork.com/joining-oin/linux-syste...


According to http://en.swpat.org/wiki/GPLv2_and_patents , GPLv2 includes an implicit patent grant

So my guess would be that the definition of "derivative work" would apply.


That's super duper grey area, which is why GPLv3 went out of it's way to have an explicit patent grant.


So the code in the linux kernel would have to be signed off/committed by MS in that case, right ? For a grant to have any meaning, the code would need to come from them ?


The definition of "Linux system" is a red herring.

If you haven't signed OIN licensing agreement, you do not receive any relevant patent grants, period.


How does that work ? OIN applies to corporations. What if you aren't a member ? Or is the actual code in the linux kernel going to have a patent grant ? But MS may not be the one committing the code. So will they just add a patent grant header irrespective of who commits the code ?


Well that's kind of shitty for the BSDs. :-/

If we took the GPL2 linux driver and imported it using our linuxkpi emulation layer, is that a "linux-derived system" enough? :)


OMG, I can't believe we finally have a cross platform read/write disk format. At last. No more Fuse. I just need to know when it will be available for my Raspberry Pi.


We already have one, it is just that few know we do:

Universal Disk Format (https://en.wikipedia.org/wiki/Universal_Disk_Format).

I can confirm it works on Linux, and works on Windows (at least W10). The wikipedia article indicates it works on newer MacOS'es, but I can't confirm that fact.


I've been using UDF formatted flash drives for a long time (10 years?) and they've been working on Windows, Linux, and Mac OS X all this time.


I thought it was for optical media only.

Yet, it seems there's no way to format to it out of box on macOS and support looks dodgy. I'm not willing to bet my data on an unreliable file system implementation.


> I thought it was for optical media only.

Nope, although it is most commonly used on optical media it is not restricted to only optical media.

From the wikipedia article:

Specifications

The UDF standard defines three file system variations, called "builds". These are:

Plain (Random Read/Write Access). This is the original format supported in all UDF revisions ... Plain build

Introduced in the first version of the standard, this format can be used on any type of disk that allows random read/write access, such as hard disks, DVD+RW and DVD-RAM media. Metadata (up to v2.50) and file data is addressed more or less directly. In writing to such a disk in this format, any physical block on the disk may be chosen for allocation of new or updated files.

Since this is the basic format, practically any operating system or file system driver claiming support for UDF should be able to read this format.


Have you used it IRL ? How does it fare performance wise ?


I've been using it for some number of years now on USB sticks that are larger than 2G (to avoid the 2G file size limit of Fat32) for file transfer to/from the work machine.

I have had zero issues with it from Linux or Win (I think I started using it when the "work pc" was Win7, but am not 100% sure of that). It has just worked, with no problems, for my use cases.

Performance wise, well, the performance is limited by the USB sticks (they are USB2 sticks) so from what I can tell, UDF is not the limiter, the USB2 interface or the flash cells in the stick are the performance limiter.


I think you still need to use Fuse for exFAT on the BSDs...

That might eventually change with the recent patent grant.


Within the constraints of the technology (e.g. >2GB files) isn't FAT32 already this? I can plug it into my Apple, Windows, and Linux devices and it will just work.


2GB is not enough and it hasn't been for a really long time.


Would have worked in 2000; nowadays, files >2 GB are an everyday occurence. (And don't even start with "but multiple-files archive"! This is not 1995, either.)


Thanks MS. Would also like to see ext4 support in Windows and better NTFS support under Linux. Then the turnaround (on the technical side at least) will be complete.


Bring btrfs to windows pls!


When will we have a modern file system that “just works” for read and write on Windows, Linux and Mac without the need to buy some proprietary product or depend on some open source option that may not be maintained or 100% reliable?

Seems like going over the network is the easiest option for people who use multiple OSes, though it comes with incompatibilities in permissions, ownership and also comes with a speed hit.

I doubt if Apple would provide implementations of APFS for non-Apple operating systems. So that probably leaves Microsoft to push NTFS (since ReFS isn’t a common option anymore as per Wikipedia) if it desires. But Microsoft seems to be on a cost cutting mode (cf the adoption of Chromium in Edge).

Are there viable and good options suitable for these three operating systems?


exFAT works on MacOS >= 10.6.5, Windows and with this announcement hopefully shortly on Linux. I'd say it counts as a modern file system too - only released in 2006.


I'd count "modern", as something like btrfs, zfs or bcahefs.


It already does in Linux albeit not by default, it needs exFAT-utils and in android Paragon has the app with license for exFAT.


Doesn't UDF work?


Last time I tried it didn’t „just work“ across operating systems, with each showing different issues.


Too little too late, with no acknowledgement of Ballmer-era exclusivity, bullying. It’s progress, but progress without awareness of baggage in the room.


"We love Linux so much that we would never let it be polluted by our horrible desktop software!"


MS wants a native kernel fs compatible with Windows so they can improve WSL. Today's virtualized kernel pays an unecessary price to access ext4 FS


Meanwhile, ntfs-3g[1] wasn't updated for years and looks like totally dead. Last version is of March 28, 2017.

[1] https://www.tuxera.com/community/open-source-ntfs-3g/


Any news on NTFS ?


NTFS-3g is pretty good and support on Linux is definitely a lot better than it used to be. Still, official (or officially supported/documented) support would be really nice.

Honestly I hope even Windows abandons NTFS soon. It's the worst part of using Windows. e.g. Try and develop any Node project on Windows and it's painful - takes literally minutes to delete the node_modules folder whereas on Linux/Mac it takes seconds. It only recently got long pathname support (260+ chars long)...


On the other side, file ACLs in NTFS are way more powerful than the Linux owner-group-world model.


You can use posix acls whenever you want though.


Posix acls aren't particularly nice though. RichACLs is what I'd love but they're not merged :(


Linux NTFS support is useful for that. It doesn't support them so you can just access/nuke any files you want through it even if Windows won't let you.


That's true of every file system if you're willing to boot into a different OS instance.


That's true. But NTFS could also improve rather than simply adopting Linux's filesystems. Heck they could improve it then Open Source the new thing.


I think ReFS was supposed to be the next-gen version of NTFS, but it never really got much use.


It's not a problem with NTFS and switching to another filesystem won't help. It's the classic "death from a thousand papercuts". The only way to mitigate performance issues is to bypass most of their I/O stack altogether, like they did with WSL2.

https://github.com/Microsoft/WSL/issues/873#issuecomment-425...


NTFS itself has had support for long paths for at least 17 years (with the prefix \\?\)


Almost every months Microsoft make a more or less major gift to the open source community, this is beautiful. They are the new "benevolent" Google and it's no surprise they're now the number one open source contributor of the world.


They already extorted Android OEMs using exFat patents for billions, might as well try to look nice.


As much as I hate the patent bullying Android should have never used FAT for ANYTHING. The stupidity around that continues to make everyone's lives hard to this day.


If you want an SD card that is also readable by the digital camera ecosystem, what choice do you have?


How often are people taking a microSD card from their camera and putting it in their phone considering on a lot of devices this means pulling the battery out.

Also they could support fat while not having the main system storage using fat.


Main system storage didn't use FAT; /sdcard did.


You need to support it if you want to properly support removable SD cards.

Microsoft managed to make their patent encumbered exFAT file system a required part of the SDXC card spec, just as their much-abused patents on FAT32 were expiring.

I guess it is gracious of them to partially undo the damage they've done, now that they have given up on Windows Phone.


Reading external flash media becomes hard if you don't have FAT support. But agreed, anything internal should have used a sane filesystem (I know that's hard to find. At least an open one: I see ext* used internally, but FAT32 for removable flash).


I always chuckled at the joke: Who makes more money off of the sale of each Android device, Google or Microsoft?


Sure sure, They had been extorting countless companies for decades including Satya's 5 years. Give me a break. They are neither benevolent nor better than Google.


That’s great but there’s nothing benevolent in their intentions. I don’t mean that as a slight, they’re the same as any other corporation - driven by profit and growth, not benevolence.


The difference is Microsoft sells and licenses products to help customers with real needs. Google's customers are all about dark marketing.


Windows Home is to Microsoft, as Chrome is to Google.

Microsoft hasn't charged consumers for upgrades to Windows for years - and surely that is because they expect to profit from advertising.

The same can pretty much be said for Windows Pro and small businesses. Why is there no subscription to Windows? Why can't I pay to get 3 more years of security fixes for Windows 7 however an enterprise can?


advertising yes, but also why they are kind of lax on piracy: the real money is in corporate, so you have to keep people using your ecosystem at home so that’s what they want to use at work


Sure, they are not the top dog anymore. They can't get away with insulting people the way they used too.

Business as usual.


They managed to overtake Apple’s market cap. Seems to me they’re quite a top dog still.



Thanks Microsoft and the developers there for sharing ExFat!

It is great to see the changed attitude and embracing of open source software! This will ease the use of large external disks, sharing files between Windows and Linux systems. It will help with larger USB memory sticks which no longer needs to be reformatted. It will also make it much easier to deal with SD cards from cameras and mobile phones.


Finally. Next steps — MS can release active sync without patents, and also start supporting Vulkan and drop DirectX 12 diversion.


wan't Samsung providing a GPL-version of exFAT for Linux already? what's the difference assuming this news means we will have a native exFAT driver in kernel "soon".

Is exFAT only good for file storage considering it has no journal ling? putting rootfs on it may have a dysfunction system relatively easily without fsck I assume.


I have extensive (bad) experience with exFAT implemntation on other platforms than Windows. My external disk formatted as exFAT gets constantly corrupted (need fsck), when used on macOS and Linux. On Linux I use this Samsung's exfat-nofuse kernel driver on raspberry pi. ExFAT fs is mounted read-only and still get corrupted.

So I don't share the excitement with most folks here. Until driver gets better, exFAT is not rock solid option for sharing data between OSes.


Samsung only had used a gpl version of exfat (mainly for android) and was forced to open source it, (or someone just did, although they did not want to admit it was gpl). This lead to development of a fuse module for linux. Which can be installed. I myself never had luck using it.


Big deal! This is right up there with Adobe Premiere on Windows finally getting a ProRes encoder at the top of my list of "Things That Would've Helped My FilmTech Career Around 6 years ago".

1) We finally have a file system that works across operating systems, which is a big deal if you ever do anything in a multi-OS environment.

2) exFAT is heavily used in the film production world, and the convenience of knowing it'll mount properly on any OS so you can duplicate it a few times is much more important than whether or not it's a "safer" file system in general. It doesn't need to survive forever -- the card just has to make it from the camera to the computer on the other side of the set, or survive a trip from one office to another on a shuttle drive that by this point is not the only place that data exists.

Regardless, it's still more mature in general than I remember it when it started to appear in high-end cameras around 2012/2013. When your Blackmagic camera shoots exFAT but isn't capable of deleting files off it without a computer...

3) Should something still go wrong (like, oh, the time I was almost responsible for losing $60k of footage my second day on a job due to unexpected use of exFAT + a truly unfortunate and odd-defying day of bad luck) the fact that Microsoft is implementing it in the kernel should still be a huge help -- it will be a proper implementation (not a reverse-engineered sometimes-working mess), will perform much better than FUSE, and just generally improve reliability. Trust me, when your card with irreplaceable footage won't mount, you'd rather not have to fight that battle on two fronts, with one of them being your Linux implementation.


fact that Microsoft is implementing it in the kernel should still be a huge help

If you're trying to recover video (which is one of the easiest types of data to recover, due to it being large, sequentially written, and bitstreams containing easily recognisable sync markers), and especially from a storage device that's being exclusively used for video, the actual filesystem doesn't really matter because practically all of the sane ones will store the file data in a contiguous range of blocks. Depending on the exact codec, you may even be able to dd the raw device into a decoder and it'll simply find the first valid sync marker and start decoding from there.


Well of course, some file carving is what has gotten me out of a few messes in the past, including the one mentioned. But in a situation where a volume won't mount/read, it's nice to eliminate a poor implementation or FUSE issue as a potential variable in determining if there's actually a problem. (Or worse, introducing a new problem... I've seen that, too, especially in earlier days of exfat-fuse/exfat-utils.)


Is it worth formatting my external flash media as exFAT at this point if I find myself dual booting often? NTFS on Ubuntu is pretty well covered so that has been what I have been using. But if exFAT support is improving I'm wondering if it is worth switching over.


Great news! But I would have delayed announcing it until I had the patches ready to be merged.


The real news here is the specification being opened.


I wonder if they are just trying to allow the exFat FUSE team to start working on the merge.


I haven't spun up anything in Azure for quite a while. Does anybody know if their default images use anything exFAT formatted? Doesn't UEFI/EFI binaries usually sit on exFAT?


By standard VFAT is required, but on most Windows PC, the UEFI partition is in NTFS.


UEFI partition is FAT32 (or EFI-FAT; EFI has their own specification of FAT, frozen in time, with it's own identifier), if not for other reason, that most UEFI implementations would not be able to boot from NTFS.

Windows installer creates several partitions: System (ESP), MSR at the beginning of the disk and Recovery at the end of the disk.


Most UEFI implementations are loaded on Intel Windows devices, which is most likely to use TianoCore, reference implementation of UEFI which includes UEFI NTFS driver by default. I have seen that some low ends from Lenovo and Acer have its UEFI partition (ESP) formatted as NTFS.


Are you sure? I had a look into edk2 and cannot find NTFS driver there. I can see the FAT one.

The NTFS support depends on device. From devices that I have around, Intel NUCs do support it, Asrock X399 board does not.

I have yet to see the Windows installer to create NTFS-based ESP partition. If manufacturer does that for specific device (where it is guaranteed, that it will support NTFS), that's different thing.


> To this end, we will be making Microsoft’s technical specification for exFAT publicly available to facilitate development of conformant, interoperable implementations.

How about their FAT patents??


Sweet. I had to purchase something for my Synology NAS fairly recently to support an ExFAT device. It’s great to see this contribution to open source.


I remember having had problems with exFAT on linux. Did it not work at all, or did you have to use user repos to get exFAT working?


You had to add it as a fuse module. I don't know if a user compiled kernel module worked. But still even if it was used as a fuse module i myself had often times problems with it. So i hope this goes away soon.


In my experience, exFAT gets corrupted constantly, and then on linux you don't have a proper fsck for it.


Will this allow WSL2 to access your entire Windows filesystem without the performance hit?

If so this is really cool.


I will believe Microsoft loves Linux when they decide to drop the development of their buggy NT kernel and starts to build windows on top of the mainline Linux kernel.

Everything else is just cheap marketing.


"buggy NT kernel" is a stretch. There's nothing particularly wrong about the NT kernel in its design or implementation, any more than other kernels.

MS don't have to drop Windows to support Linux. They can walk and chew gum at the same time.


Does this mean it is royalty free now?


embrace, extend, exFAT


So after abusing this bullshit for more than a decade to bully companies. How gracious and generous of them. Oh btw, Satya is the CEO for 5 years.


Stop shitting on them. We won the war. They’re mending their ways. It takes a long time to right such a large ship with levels upon levels of bureaucratic bullshit that’s been caked on for decades.

It’s a WIP and they just granted a patent license to the OIN for fuck’s sake. Even 10 years ago this would have sounded like science fiction.


The WSL looks a lot like the embrace step of an embrace-extend-extinguish doctrine. I'll believe them more when I can write user-facing apps against their posix subsystem instead of requiring people to jump through many many hoops to actually install it.

Right now it just looks like something they were forced into doing because they were losing on the devtools market.


WSL 2 foregoes the system call translation approach in favor of virtualization, you get a Linux kernel: https://devblogs.microsoft.com/commandline/announcing-wsl-2/


>instead of requiring people to jump through many many hoops to actually install it

Yeah the hoops of going to the Store, searching for "ubuntu" and clicking Install


Are they not still requiring this?

    Enable-WindowsOptionalFeature -Online -FeatureName Microsoft-Windows-Subsystem-Linux
I admit I haven't used WSL in a while.


You still need to enable the subsystem, but I don't see why most users need it on by default. It seems like an easy enough feature to prompt the user to enable (or enable for them automatically) if your application calls for it.


Programs and features control panel, then add/remove windows features, or through command line.

The exact same thing they require to enable or disable any optional windows feature like ie 11, windows media player, I is, ...


> Stop shitting on them. We won the war. They’re mending their ways.

It aligns with the current management that they're doing what they're doing. They had "terminal services for Unix" for some time, then there was war on Linux, now it's open standards again, and it may change again in the future.

"We" didn't "win" annoying. It's just a happy period. Let's be happy about it and not take it for more than it is - liking what the current MS management does.


Didn't we just have a post complaining that the Microsoft Skype web client rejects Firefox, for no obvious reason? Yes, it was https://news.ycombinator.com/item?id=20818750.

Maybe their interests have just shifted higher on the protocol stack.


No, we are losing the war.

Open Source is not an impediment to Microsoft's attacks on users' freedom.

You can get a job on LinkedIn to write code on GitHub that runs on Windows and talks to a server deployed to Azure with a single click from within the README.md .

In that ecosystem, permissively licensed Open Source software just helps Microsoft exploit you.

Which is why Microsoft unironically love Open Source and are very happy to see paid and unpaid shills shout that from the rooftops.

Dumping soon-to-expire patents to get goodwill from people they need to encourage into or prevent leaving that ecosystem is a strategic move that is designed to make sure you don't escape in the long term.

Pointing this out isn't shitting on the poor, defenceless corporation.

It's just not being fooled by their attempts to fleshmask Open Source.

They still hate Free Software.


This is not just a dump of soon to expire patent. The patent grant and their support and commitment to indemnify anyone who uses it through OIN proves to me, that at least when it comes to this, they are acting in good faith.

Listen, my first job outta high school was at Red Hat. For the majority of my life I hated Microsoft and everything they stood for. The bottom line is, money talks (the virtues or lack thereof of that are for a different debate) and they realized that people became familiar with Linux which was superior for most non-desktop use cases, enabled innovation at a greater pace and also was a path around all the Microsoft bullshit and customers were abandoning ship. Eventually they had NO CHOICE but to evolve or die.


> Eventually they had NO CHOICE > but to evolve or die.

So, really they haven't done anyone any favours then?

Why do you think so many people look upon their acts as something with good intentions?

Just better marketing spin?


I'm not so sure "we" won the war. If anything it's more like they mostly won the war (for the Enterprise Desktop, which is where the real money continues to be for them) and ALSO found a new SAAS/cloud empire and now they need to win the peace.


"now they need to win the peace."

Or the next war


So if you're not party to the OIN, are you allowed to use / implement exFAT? Or are you still blocked by the patent?


> Stop shitting on them.

.... no?

> We won the war. They’re mending their ways.

"We" (?) are either there or not. The half-measures of mending continue at the pace of their choosing.

> Even 10 years ago this would have sounded like science fiction.

A strong point in favor of continued skepticism.

Nobody is calling Red Hat's Ansible Tower or other closed-source products some change of heart in the company and this is no exception in the reverse.

edit: Always good to be downvoted when responding to a comment poisoning the well (and cursing) with a comment that does neither.


I think that's pretty unfair - Microsoft has changed enormously in the past 5 years.


MS has changed from proprietary software vendor to a spy-on-the-users-and-try-to-monetise-them services company that's trying to be Google. I don't think that's much of an improvement.

Their code used to be their core value and source of profit; they are open-sourcing a lot of it now because that's not true anymore and they might as well get some publicity out of it. If you look at companies like Google and Facebook who publish lots of open-source, you'll see the same trend --- it's opened because it's not worth anything to keep it secret anymore.


VSCode, exFat, and .Net core are not enormous. Their revenue streams are enormous and remain as close to their chest as ever.


So they should start losing money? Should Google? Should Apple? I don't get this argument. And .Net Core being open source is enormous if you use it.


The system I just implemented is using .Net Core on Ubuntu on AWS/VMWare and is happily processing lots of transit trips a day. Licensing costs are $0.

.Net Core brings MS's equivalent to Java to Linux and enables the use of a better language set (C#/F#) and allows for a relatively seamless development experience.

What's not to like?


>So they should start losing money?

I was unclear; their revenue streams are still overwhelmingly comprised of proprietary tech.

Darwin and AOSP are massive compared to the toys MSFT is throwing over the Linux wall. This idea that they need to lose money is a strawman.


I disagree - VSCode has an enormous user base, and .NET Core is a huge deal. They even open sourced the serverless engine from Azure's Function Apps!


All that withstanding, nobody but developers buy Microsoft products because of VSCode.


Microsoft SQL Server on Linux indeed has enormous licensing costs.


Relative to what, though? Departmental budgets or MSFT's gross revenue?


Hmm.. Oops! now exFAT will bring the [A-Za-z] case in-sensitivities into the GNU/Linux system.


The ext4 folks are working on adding case-insensitive support... https://lwn.net/Articles/784041/




Consider applying for YC's W25 batch! Applications are open till Nov 12.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: