Will Floating Point Textures Be Merged Into Mesa?

[yotambien]

I think it's 2020 or so. Quite a while. And I'm also pretty sure these patents are registered in Europe, so everyone who thinks this is a US only problem is wrong. I know some of the opengl patents are registered there anyway, if these aren't.

It isn't whether the patents are registered or not what's important. What's important is that European law doesn't (in principle) recognise software patents. That the European Patent Office grants those patents is irrelevant, the courts decide when the time comes.

[RealNC]

Not sure where, but I remember reading somewhere that the Europe vs USA patent thing basically boils down to fairness and that in Europe software patents exist and a court recognizes them as valid, but can still decide not to apply them on a case by case basis without actually invalidating the patent in question.

[HokTar]

Do you watch SKY?
Do you buy DRMed media(virtually any offline store)?
Do you visit cinema?
Do you buy windows games from famous publishers?
Do you buy proprietary software?

Because, even if you just copy them yourself, you contribute to market statistic.

Have you played Star Control II? Check out who Druuge are.

Actually I do _none_ of the above, believe it or not.

[Jonno]

Not sure where, but I remember reading somewhere that the Europe vs USA patent thing basically boils down to fairness and that in Europe software patents exist and a court recognizes them as valid, but can still decide not to apply them on a case by case basis without actually invalidating the patent in question.

Not quite, the issue is that the EU Directive on patents (European Patent Convention) explicitly excludes "programs for computers" from patentability, but the European Patent Office has no political or judicial oversight and the bureaucrats working there can thus award any patents they want to with no repercussions.

However, as most member countries do follow the European Patent Convention in their national laws, and judges generally do follow the law, these granted patents are useless for anything but FUD in most member countries.

One notable exception is Germany, that violates the European Patent Convention and does allow for software patents. Fortunately the German legal system does allow for judges to use common sense in patent cases, so the damage done is limited.

In theory the European Commission could sue the German government for the violation and force them to pay a large biannual fine until Germany either complies with the directive or secede from the union. However, that is not likely to happen, as the European Commission are the only one with standing to sue, and they have tried, twice, to introduce a EU Directive that would mandate software patents in all member countries.

There might be other countries whose laws violates the European Patent Convention, but if so I'm not aware of them.

As for "not to apply them [...] without actually invalidating the patent", that is the only thing a court can do. The only ones that can invalidate a European Patent is the European Patent Office. As I said, there is no judicial oversight. And the precedent set by one court case is only valid in that jurisdiction, so another court might think differently.

[Qaridarium]

Not quite, the issue is that the EU Directive on patents (European Patent Convention) explicitly excludes "programs for computers" from patentability, but the European Patent Office has no political or judicial oversight and the bureaucrats working there can thus award any patents they want to with no repercussions.

However, as most member countries do follow the European Patent Convention in their national laws, and judges generally do follow the law, these granted patents are useless for anything but FUD in most member countries.

One notable exception is Germany, that violates the European Patent Convention and does allow for software patents. Fortunately the German legal system does allow for judges to use common sense in patent cases, so the damage done is limited.

In theory the European Commission could sue the German government for the violation and force them to pay a large biannual fine until Germany either complies with the directive or secede from the union. However, that is not likely to happen, as the European Commission are the only one with standing to sue, and they have tried, twice, to introduce a EU Directive that would mandate software patents in all member countries.

There might be other countries whose laws violates the European Patent Convention, but if so I'm not aware of them.

As for "not to apply them [...] without actually invalidating the patent", that is the only thing a court can do. The only ones that can invalidate a European Patent is the European Patent Office. As I said, there is no judicial oversight. And the precedent set by one court case is only valid in that jurisdiction, so another court might think differently.

you are wrong in germany there are the same patent law as in EU and the EU+germany allow software parents if the pice of software act with hardware means an software binding to an pice of hardware .

exapel for that is fat32,s3tc and so one.

you can not patent a pure software allgoritm or software only the spezific interaction with an pice of hardware,.

and in this point of view there is no different between EU and germany.

[Jonno]

you are wrong in germany there are the same patent law as in EU and the EU+germany allow software parents if the pice of software act with hardware means an software binding to an pice of hardware .

exapel for that is fat32,s3tc and so one.

you can not patent a pure software allgoritm or software only the spezific interaction with an pice of hardware,.

and in this point of view there is no different between EU and germany.

That is the fringe logic of the European Patent Office ("we don't grant a patent on the software algorithm, we are granting a patent on running the software algorithm on a general purpose computer") that has no basis in any EU Directive or the law of most member countries (Germany excluded).

In Sweden (and most of EU) you couldn't patent the s3tc algorithm running on a computer. You could possibly patent a particular way of implementing it in hardware, but never a software implementation, including using general purpose shaders on a GPU, and any other way of implementing it in hardware would be fine.

And the vfat patent is in no way tied to the HDD, it covers how store an alternative (long) file name in a directory structure without breaking compatibility with the old MS DOS implementation that only understand the original (8.3) file name. The only hardware it is "tied" to is the general purpose CPU that executes the code.

While all patents granted by the European Patent Office technically is valid in the whole of the European Union, trying to sue anyone over the two patents you mentioned would be laughed out of court in any member country whose laws actually follow the European Patent Convention. At least if the defence manages to explain to the (possibly computer illiterate) judge what is going on.

[Remco]

That is the fringe logic of the European Patent Office ("we don't grant a patent on the software algorithm, we are granting a patent on running the software algorithm on a general purpose computer") that has no basis in any EU Directive or the law of most member countries (Germany excluded).

I think the easiest to understand argument against this faulty logic is that you don't actually need any hardware when it comes to illegal software patents. You can simply calculate the result of the 'patented' equation on a whiteboard in math class. You're the general purpose processor executing a mathematical function. And math is not patentable. A legitimate patent that describes a particular hardware design does not have this kind of argument against it. Then it is about the hardware, not the math.

[DanL]

This won't be a problem for us Gentoo users.

It won't be a problem for anyone who knows how to compile software or add a repository to their sources. SHUT UP you Gentoo elitists

[Prescience500]

I think merging this in is a good thing. It'll make it easier to maintain and eventually add it to default mesa builds when the patent/s on it either expire or get thrown out by courts (or legislative bodies) or an FOSS driver on a FOSS operating system exception is given.

[myxal]

have it only built if during the configure process an argument like --enable-patented.

Hear hear, a voice of reason!

It's about time maintainers and distros wake up to the fact there's a whole world beyond countries upholding software patents, and neither developers, nor users want to be constrained by stupid legislation. This MO worked fine for freetype, I'm sure it will work elsewhere where distros face similar problems.