Linux Support For Microsoft's exFAT File-System

[rzrx]

the non-fuse exFat kernel module has been released and available here: https://github.com/rxrz/exfat-nofuse
tested on 3.8.11, needs fixes to work with kernel 3.9+

[mirza]

A spec is useless if code is never created to benefit the large majority of users out there and can never be expected to become mainstream catering to a small niche group of users.

Do you really not see that Microsoft NEVER pushed common standards if there was any chance of pushing own propriatery patent-encumbered alternatives? Yes, they support TCP/IP, HTML, maybe few others, but only when they are late to the game and had to support it or customers would bitch/go away. Otherwise: ODF -> OOXML, Java -> .NET, OpenGL -> DirectX. So it is *always* non-standard, closed specs, patent encumbered replacement. Why this FS is supposed to be any different? It isnt, they created again own propriatery undocumneted implementation of FS behind closed doors without any effort to talk to anyone else.

[deanjo]

How many "other OS's" are there? I'm writing a micro-kernel myself. Can I ask FS-code writers to write a port of the FS for my OS? You can NEVER port something to an infinite number of OSes. All you can do (and should do) is to fight for open specifications, not open source. Anything that is (Windows + Linux)-specific or (Windows + Linux + Mac)-specific is just as bad as being Windows-specific or Linux-specific. The only FSs that deserve to be called as "open filesystems" are those that have open specifications.

A spec is useless if code is never created to benefit the large majority of users out there and can never be expected to become mainstream catering to a small niche group of users.

[BhaKi]

Hindered or not, if the FS does not have a port to other OS's it's just as useless outside of it's native OS.

How many "other OS's" are there? I'm writing a micro-kernel myself. Can I ask FS-code writers to write a port of the FS for my OS? You can NEVER port something to an infinite number of OSes. All you can do (and should do) is to fight for open specifications, not open source. Anything that is (Windows + Linux)-specific or (Windows + Linux + Mac)-specific is just as bad as being Windows-specific or Linux-specific. The only FSs that deserve to be called as "open filesystems" are those that have open specifications.

[deanjo]

The Linux/BSD-community already came up with plenty of alternatives, e.g. the very lightweight ext2.

No they didn't they came up with a solution for their OS. A real cross platform filesystem that was designed with maximum compatibility is like UDF or ISO 9660.

But, and here is why politics matter, no file system will ever have a chance to become the successor of FAT32 as the interoperable file system without Microsoft's blessing. They still hold an overwhelming percentage of the market.

Again this is simply not true as it has been demonstrated with UDF ans ISO-9660. These specs were not designed by MS but yet all OS's adopted them including MS.

In other words: what they distribute with Windows becomes a defacto standard, especially with something like a file system.

Again this isn't necessarily the case there have been many cases where even though the industry may support some of MS's proprietary formats they do use others as the industry standard. Take a look at the codecs, MS has been pushing WMV / ASF for years but those are not considered the standard.

Most Windows users have no idea what a file system is or that their computer has got something like that. They won't go to fs-driver.org to download an ext2 driver.

Sure they will, if a device needs supporting software that is exactly what they do now. People download drivers, codecs, etc all the time.

[BhaKi]

Booo Hooo, I don't see other file systems bending over backwards to accommodate other OS's as well. So MS makes another FS, so what. I don't see the opensource community making their file systems easily accessible on windows, os x, insert alternative OS name here. If you run 64-bit windows the only option you have is a slow ext driver, Perhaps if the foss community would put some effort into making their FS easily accessible in windows you wouldn't have to worry about stuff like this. I only can dream so far of being to easily r/w to filesystems like btrfs, xfs, etc in windows.

The problem is not about source code as much as specifications. NTFS, EXT2, EXT3, exFAT are all crap. FAT is the only popular file-system to have a public specification. So any OS-writer can add FAT-support to his/her OS by reading those specs. Ext2/3 are not so open. I guess that's because people are so happy to equate open source to open specs.

[some-guy]

There's already JFFS2 and YAFFS, but of course, M$ is always right...

[piquadrat]

No, I'm saying instead of bitching what people can't change, do what they can do and that is come up with a viable alternative supported on all platforms and stop worrying about what some corp is doing on their product. When it comes to things like filesystems there should be less politics

The Linux/BSD-community already came up with plenty of alternatives, e.g. the very lightweight ext2. But, and here is why politics matter, no file system will ever have a chance to become the successor of FAT32 as the interoperable file system without Microsoft's blessing. They still hold an overwhelming percentage of the market.

In other words: what they distribute with Windows becomes a defacto standard, especially with something like a file system. Most Windows users have no idea what a file system is or that their computer has got something like that. They won't go to fs-driver.org to download an ext2 driver.

[chaos386]

The same as usual, the companies should stay with Fat32 and not accept this crap of FS.

Also, exFat didn't have advantages compared to NFTS or Fat32.
http://www.testfreaks.com/blog/infor...ntfs-vs-exfat/

The problem with Fat32 is that files can't be larger than 4 GB, which is a serious restriction when you're, say, a consumer electronics company developing an HD camcorder that uses flash cards for storage. File systems like NTFS, EXT, XFS, etc. have too much overhead and are way too hefty to be practical (why would a camcorder need file permissions, journaling, or any other advanced feature?). They wouldn't add anything of value but would require larger, more expensive and power-hungry chips, the firmware would be more complex, and you'd have to hire new firmware engineers who know the new file system.

This might explain why the SD Association didn't go with something like UDF: exFAT seems to just be an extension of FAT32, so it probably doesn't take much work to modify existing products to use it.

[deanjo]

A
In other words, why does Linux always have to support other stuff, while where other OSs (read: Microsoft) need support for FOSS projects, the FOSS community gets blamed?

Because they are the ones bitching.