Merging exFAT Support For Linux Is Being Talked About - Waiting On Microsoft's Blessing

[shmerl]

Even if they seem friendly, it could be bait and switch. Who knows what will happen with Microsoft in the future? What if the next CEO will try to abuse any patent they have to crush competition?

They'd need to quit OIN then.

[carewolf]

The code already exists. But patents are different from copyright as a clean room implementation of the features will still infringe the patents.

Yep, even if you invent something completely on your own in garage without ever hearing about other people doing anything similar. If somebody else patented it first, you have to pay them..

That is how patents "work".

[anarki2]

About time.

[phoenk]

It's not supposed to. A patent is supposed to cover one specific way of implementing something. If you come up with a different way of doing the same thing (a clean-room implementation, for example), with even 1 step being different, then it's not supposed to be infringing. In fact, if you patent an application and describe it with C++ code, someone implementing it in Java wouldn't (or shouldn't if things were done right) infringe.

However, the way the US patent system has been subjugated over the past decades allowing less and less specific patents, covering any possible implementation of a series of generic steps, it would probably be hard to convince a judge of that.

You're partially right. A patent covers a particular way of doing something, in this case storing and reading data on a disk. The code to do this has copyright, and at least one of those implementations is open source. However, a filesystem is a specific way of putting data on the disk, and that is what is patented. Anything that would be compatible with the other exFAT drivers would have to use at least some of the patents.

[Jaxad0127]

Yep, even if you invent something completely on your own in garage without ever hearing about other people doing anything similar. If somebody else patented it first, you have to pay them..

That is how patents "work".

That only happens if you manage to do something sufficient identical to the patent. In general, there are many ways to do things, so that isn't an issue.

They way software patents are currently handled does have that issue. Other (more traditional) areas don't (as far as I know).

[smitty3268]

It's not supposed to. A patent is supposed to cover one specific way of implementing something. If you come up with a different way of doing the same thing (a clean-room implementation, for example), with even 1 step being different, then it's not supposed to be infringing. In fact, if you patent an application and describe it with C++ code, someone implementing it in Java wouldn't (or shouldn't if things were done right) infringe.

That doesn't have anything to do with reverse-engineering, though. You just have to come up with a different way of accomplishing the same thing, and usually basing it off the original solution is actually the easiest and fastest way of ensuring that you're doing that. A clean-room implementation can very easily just end up doing things the same way as the patent, since that likely covers at least the most obvious and easy way of accomplishing it.

Also, patents are never language specific. They have pseudo-code at best, which would work across languages, and any lawyer will try their best to avoid even that and make everything as generic and meaningless as possible.

Something as specific as syntax would only be covered by copyright - which is indeed what clean-room reverse engineering is meant to work around.

[nivedita]

It's not supposed to. A patent is supposed to cover one specific way of implementing something. If you come up with a different way of doing the same thing (a clean-room implementation, for example), with even 1 step being different, then it's not supposed to be infringing. In fact, if you patent an application and describe it with C++ code, someone implementing it in Java wouldn't (or shouldn't if things were done right) infringe.

However, the way the US patent system has been subjugated over the past decades allowing less and less specific patents, covering any possible implementation of a series of generic steps, it would probably be hard to convince a judge of that.

You’re confusing patents with copyrights.

[Guest]

That doesn't have anything to do with reverse-engineering, though. You just have to come up with a different way of accomplishing the same thing, and usually basing it off the original solution is actually the easiest and fastest way of ensuring that you're doing that. A clean-room implementation can very easily just end up doing things the same way as the patent, since that likely covers at least the most obvious and easy way of accomplishing it.

Also, patents are never language specific. They have pseudo-code at best, which would work across languages, and any lawyer will try their best to avoid even that and make everything as generic and meaningless as possible.

Something as specific as syntax would only be covered by copyright - which is indeed what clean-room reverse engineering is meant to work around.

Well. The patent could just cover the mechanism how (meta) data is stored on disc. Then any implementation using that (meta) data storing mechanism would infringe the patent.

To not infringe such patent, a different or modified data store mechanism, meta data representation, would be needed. Therfore, it wouldn't be compatible 100% with exFAT.

So. There's no way to accomplish same thing, if M$ was not dumb enough to patent pseudo code, but instead they patented (meta) data store mechanism on HDD.

[Guest]

The fundamental issue with Microsoft's proprietary disk formats is just that, they're proprietary. Even if you figure out a completely legal way of reverse engineering it and creating an alternate fully compatible implementation akin to how the original IBM PC BIOS was cloned, Microsoft is bound to follow IBM's example and sue anyone brave enough to actually do that. Microsoft's bound to lose it if their victim decides to really fight the lawsuit, but the cost of doing so will be enormous and it's incredibly rare to have your legal expenses reimbursed after a spurious lawsuit. Even when it's plainly obvious.

Try to remember that Microsoft has a long history of being very hostile towards open source competitors with how they funded the SCO lawsuits, how they've extracted patent royalty fees from Android device makers and more recently how they left the exFAT patents out of the opening of their patents to open source efforts.

I'm really not kidding about them a history of good ol' fashion patent racketeering with android device makers. They even had the gall of running a campaign to android device makers about how Android "isn't free" when they have to pay them royalties for things like FAT/exFAT support and promising them to waive those royalty fees if they also made Windows Mobile devices. Needless to say not much came out of those few Windows mobile devices that primarily HTC put out so they didn't have to pay Microsoft royalties for their use of Android. (As for Nokia, before they literally bought the company's phone division they literally paid them to switch to Windows Mobile)

Yeah, IIRC Barnes and Noble was the only company that fought back and I believe all of the patents in question were shown to be invalid, as a result. It does look like Microsoft has changed considerably since those times, I'm hoping they don't repeat that kind of nonsense again.

[Pajn]

It's not supposed to. A patent is supposed to cover one specific way of implementing something. If you come up with a different way of doing the same thing (a clean-room implementation, for example), with even 1 step being different, then it's not supposed to be infringing. In fact, if you patent an application and describe it with C++ code, someone implementing it in Java wouldn't (or shouldn't if things were done right) infringe.

However, the way the US patent system has been subjugated over the past decades allowing less and less specific patents, covering any possible implementation of a series of generic steps, it would probably be hard to convince a judge of that.

Normal patents yes, but for software that is already covered by copyright. Software patents are therefore very generalized, hand wavy description of the theoretical structure and not the specific algorithm. For example, the Hindley Milner type system (six basic equations) would be covered by a software patent, not the W algorithm implementing it.

For that reason, democratic countries with a functioning justice system does not allow software patents to exist. But due to a few dysfunctional countries having huge economic power, we all have to live with this crap.