Samsung Properly Open-Sources exFAT File-System

[ArneBab]

Yep, looks like Samsung forgot to remove the "or any later version" part. Looks like a mistake, since the Linux kernel itself doesn't require it.

I don?t think they just forgot it, that would be too big an oversight. It could be part of the agreement with the conservancy, though - but both won?t be talking about that

[RealNC]

Besides: GPLv2 already contains an implicit patent grant, so Samsung had better work this out ? with the explicit GPL release (GPLv2 or later)

Yep, looks like Samsung forgot to remove the "or any later version" part. Looks like a mistake, since the Linux kernel itself doesn't require it.

[ArneBab]

Yay GPL!

Say what you want: GPL works. The Alternative for Samsung would have been to lose all rights to distribute Android.

So, how do you think this would have gone with a non-copyleft license?

Besides: GPLv2 already contains an implicit patent grant, so Samsung had better work this out ? with the explicit GPL release (GPLv2 or later), they are obliged to ensure that downstream recipients aren?t restricted - as long as Samsung distributes the code?. See http://en.swpat.org/wiki/GPLv2_and_patents

?: For Samsung this would at least be damage-minimization (of their own infringement): If Microsoft starts the patent-war, Samsung won?t lose the right to all GPL code in Linux, just the right to distribute this part (and the code is officially out).

[mrugiero]

People seem to keep saying this, but it's not true. If you distribute GPLv2 material commercially, you have two choices. You can distribute the source code alongside the binaries (section 3(a)), or you can include an offer to provide the source code on request (section 3(b)). 3(a) means you only have to provide source code to people you ship the binaries to, but nobody ships products like this - you'd need to include an extra DVD in the box, and basically nobody would care. So everyone ships under 3(b), and GPLv2 clearly states:



the key being "any third party", which includes people you didn't give the binary to. Samsung didn't include a copy of the source code with the device, therefore Samsung were distributing under GPLv2 3(b), therefore Samsung has obligations to the entire world.

Also, even if Samsung doesn't have such obligation with the entire world, Samsung must give it to whoever they distribute binaries to under the GPL, and that means anyone receiving it is free to share it, or merge it in a GPL compatible project.

[mrugiero]

The reason kernel modules were designed in the first place was to circumvent licensing issues. But this case is probably even worse.

Do you have any historical proof about that? Because the intuitive reasons are avoiding the need to recompile the kernel when programming on non-core pieces, and the ability to distribute smaller packages for drivers instead of the whole kernel (and to avoid having multiple, different kernel images or a huge one including everything).

[GreatEmerald]

A camera shooting ogv or VP8 video onto an ext4 camera card with custom firmware would be damned fine with me. Ideally, you would be able to install free software on every device you own and thus blacklist every last piece of patented software.

In other words, a camera that shoots Lib-Ray directly? Alternatively, if you really need that Windows compatibility, UDF could be used as the base file system.

[Luke]

Distros are easily kept out of the lawsuit line of fire

To clarify my point: It's a driver, and a kernel driver, not a user-land one, at that. The one to get sued for it's use would be the distributions and the big corporate backers and users, not the individual end users.
The reason kernel modules were designed in the first place was to circumvent licensing issues. But this case is probably even worse.

Protecting distros is simple: No distros install it by default, at worst none have in in any repo hosted in a vulnerable location, nor under their own name. You can't sue Debian because somewhere else there's a repo anyone can add to Debian containing libdvdcss. Libdvdcss, in fact, is an example of how to handle an especially "hot" piece of software. Nobody's been able to get it off the whole internet, and nobody's bothered for using it if they are not a coporation vulnerable to external inspection (by an employee,fired for trying to organize a union perhaps).

Hell, you could put Libdvdcss, all the codecs, even Flash and prop drivers in Trisquel if you really wanted to, and a lawsuit against Trisquel because someone put libdvdcss or ExFat into their system based on it would probably be laughed out of court.

Big corporate users are being advised by Mint and other such distros not to install codecs either. Only someone using a big still camera with an unpatented raw format for everything will need ExFAT but not need the patent-busting codecs just to read the camera files. As a result, the importance of the ExFat patent is limited by the fact that patent-busting codecs are in most of the same workflows that need the ExFat driver.

There remains a reason to boycott cameras that support ExFat at all: to deny Microsoft revenue from the license purchased by the maker of the camera. Make sure it does not support camera cards over 32GB, just use "seamless recording' to start new files every 4GB-it works fine.

Also, once we have to use external repo kernel modules to handle "jerk" fileystems like Exfat, every kernel update is going to require a DKMS run, but that only slows down kernel updates, no every single read of a camera card. It's probably too late to kill ExFat commercially, the real fix would be custom firmware or OS images for common cameras that would use our own filesystems-and our own codecs. A camera shooting ogv or VP8 video onto an ext4 camera card with custom firmware would be damned fine with me. Ideally, you would be able to install free software on every device you own and thus blacklist every last piece of patented software.

[Bernard Swiss]

Fuck Microsoft, and their patents. It is wrong to patent something like a filesystem in the first place, to hinder interoperability with free software. Especially since it's not especially innovative, it's of poorer quality than FAT32. It supports larger volumes, larger file sizes and has some arbitrary limits removed, but has very poor fault tolerance. (not even any backup copies of the file allocation table)

Does the invalidation of earlier FAT related patent have any bearing on ExFAT patents?

[curaga]

Michael, benchmarks of this coming up?

[mjg59]

Regarding 3, the obligations of the GPL only applies to the users you are distributing the binaries to.
Samsung only distributes the binaries on ExFAT licensed devices, so the users have the right to modify Samsung's code AND use it (and, with GPL, have the right to redistribute the code).
Samsung has zero obligations to the rest of the world.

People seem to keep saying this, but it's not true. If you distribute GPLv2 material commercially, you have two choices. You can distribute the source code alongside the binaries (section 3(a)), or you can include an offer to provide the source code on request (section 3(b)). 3(a) means you only have to provide source code to people you ship the binaries to, but nobody ships products like this - you'd need to include an extra DVD in the box, and basically nobody would care. So everyone ships under 3(b), and GPLv2 clearly states:

Accompany it with a written offer, valid for at least three years, to give any third party, for a charge no more than your cost of physically performing source distribution, a complete machine-readable copy of the corresponding source code

the key being "any third party", which includes people you didn't give the binary to. Samsung didn't include a copy of the source code with the device, therefore Samsung were distributing under GPLv2 3(b), therefore Samsung has obligations to the entire world.