I've been involved in an open source project (FreeBSD) for a long time. There have been a number of complaints about GPL violations in the past. These complaints are usually made in private. That helps a lot. Often times the compaints are wrong (The GPL code that was alleged to have been taken and improperly included in FreeBSD turned out to have been taken from BSD 4.4lite and incorporated into the GPL code was the worst example). There have also been cases where the same code appeared in drivers in multiple places. Again, that wasn't a GPL violation because both places took the code from a common data sheet. Sometimes supposed violations are cleaned up out of an abundance of caution: it isn't clear the code is improperly included, but the code in question is easy to rewrite and/or icky to start with.

There are also times where GPL code is improperly imported code from BSD as well. Even when these are found it isn't always worth it to complain. Sometimes the gain from complaining is so small that it is easier to just let the folks use the code and not worry too much about it. Sometimes having the code out there and improperly licensed is better than getting it removed from the code base.

In general, I've found that most people that aren't lawyers don't know the law or the provenance of the code very well. By complaining in private, you get a chance to learn a bit about both. You also give people a chance to make it right. With large open source projects, the chances for accidental mistakes are high. The projects are generally keen to avoid the mistakes in the first place, and even keener on making sure that they get ironed out after the facts. Turns out most companies have a similar view and will do the right thing when asked (but sometimes it takes a little time, which is OK: the GPL never said instantly on demand).

Of course, this begs the question about the validity of the License to use GPL software after a violation has occurred, the scope to which license is lost, how to get it back, etc. GPLv2 is silent on the issue, while GPLv3 gives you one shot to fix it (but that's likely insufficient for large companies that have multiple product lines done by disjoint sets of people all of whom aren't educated on the finer points of incorporating GPL software into their products).

The GPL has been held to be valid a number of times in a court of law in different countries. This is true. However, the lawyers didn't say it was invalid, so it is also irrelevant.

They said that it was unclear what is meant by derived work, and therefore it was unclear what could be licensed with a different license when combined with GPLv2 software, and what had to have a GPLv2 license. It all hinges, according to them, on if one takes an expansive view or a narrow view reading of the independent work clause. This is something that's very much up in the air right now, with many people playing fast and loose with the rules. You have a continuum of behavior here. Everything from "I wrote this file, therefore I don't have to license it under the GPL, even though it is linked into the kernel" to "I GPL'd the shims to my proprietary driver, but not the driver itself." The authors point out it is unclear how much of this behavior is safe and how much isn't. The ambiguity and shifting attitudes about what is and is not a derived work creates risk and uncertainty when using this license.

They claim GPLv3 doesn't suffer from these weaknesses.

Nowhere to do they claim the GPLv2 is not legally valid. Just that ambiguity exists,

This assumes that those binaries are recompiled.

assert(sizeof(foo) == sizeof(bar)) (or the compile time variations) is just as effective...

I had a Libretto 50CT, which pre-dated the 70CT. I loved the size (it was almost exactly the same size as a VCR tape) for portability. Had a slow Pentium processor in it. I dropped mine and the warrantee couldn't fix it so it was replaced by a Sony that I didn't like as much.

The keyboard was small, and kinda hard to type on. But I got used to it. I did a lot of development on FreeBSD PC Card and CardBus stacks on that little box. I do miss it, except when I need to see a lot of data on the screen. Then I like my newer laptops better...

There was also this crazy libretto mailing list for hacking the suckers. People posted how to overclock them by soldering and removing 0 ohm resistors, how to build car power supplies (I built one and learned a lot), how to add brightness enhancing films, how to hook up better microphones, etc. These things found their way into lots of small environments before the Soekris boxes became popular for such things.

But having used it, I do know what the limitations of the new crop of netbooks have. They are kinda cool, and all run FreeBSD very well, but I haven't jumped in yet... My life has changed a lot since I had the Libretto and I'm no longer sure it is a good fit. I like the bigger text on my "newer" laptop, but miss all the quirkiness of the Libretto....