OSNews

In order to distribute someone else's code under GPLv2, you must provide a royalty-free license for any of your patents which affect that code.

For example, I release some code under the GPLv2. You use my code and modify it in a way that makes it subject to your patent. You only have my permission to distribute *your* code, which includes *my* code, if you conform to the terms of the GPLv2. Those terms include granting the free use of your patent to anyone who obtains your code under the GPLv2. If you want to charge royalties for your patent, then you can't distribute my code. If you want to charge royalties to people who aren't using the GPLv2 code, go right ahead.

DRM is a broad subject. Using crypto techniques to control what may execute on a computer is a useful technique. The problem is who decides what's allowed. My belief is that DRM is good when controled by the owner of the computer, and bad when it takes control away from the owner and gives it to someone else. The RIAA/MPAA want that control. Once more, control isn't the problem, it's who has control.

Nothing about DRM prevents merging branches back together. It may prevent *running* the merged code on a particular piece of hardware. But the code is available to everyone, and may be used elsewhere. That's what Linus cares about - changes to the code are available to anyone who wants them. If you buy hardware that won't run certain code, that's an issue with the hardware, not the license of the software. If you don't like closed hardware, and I don't, then don't buy closed hardware. But closed hardware does not prevent the software from evolving, which is the most useful feature of the GPLv2.