Unfortunately the FSF thinks the GPL end justifies any misleading means [LWN.net]

Unfortunately the FSF thinks the GPL end justifies any misleading means

Posted Sep 11, 2010 15:22 UTC (Sat) by coriordan (guest, #7544)
In reply to: Unfortunately the FSF thinks the GPL end justifies any misleading means by FlorianMueller
Parent article: FSF responds to Oracle v. Google and the threat of software patents

FSF said that the implied patent grant works in the USA but that the problem was that it mightn't work in other countries. (Eben doesn't usually speak for FSF but in that 2006 presentation he was presenting an FSF project.)

I think I remember RMS giving the UK as an example of where it mightn't work.

The Oracle v. Google case is in the USA, but the European Commission's comment is interesting. They did confirm that an implied grant existed, even if they put narrower limits on it than we'd like.

It seems clear to me that if you take OpenJDK (with features A, B, C) and you add more code (features X, Y, Z), then all downstream versions (including through an intermediary such as IcedTea) have a patent grant for features A, B, and C. And that's in the narrow interpretation of the EU.

Maybe it'd be possible to build something like Dalvik while staying within those limits - I don't know, it would depend on how many of the changes are additions, how many are changes to the existing code, and whether those changes are "substantial" in the eyes of a judge. If the USA has a broader view of implied patent grants, then it's more likely that Google would have been safe in the USA.

(I'm surprised by Dan Ravicher's comments. I think he's said elsewhere that the implied grant does work. I wonder if the second-hand reporting of his statements changed them slightly.)

(Log in to post comments)

Unfortunately the FSF thinks the GPL end justifies any misleading means

Posted Sep 11, 2010 15:31 UTC (Sat) by FlorianMueller (guest, #32048) [Link]

Thanks. Please note two things about what the European Commission decision on Oracle-Sun says in this context:

1) They defined relational database management systems as a worldwide market for purposes of competition law, so they were also interested in the situation under US law, although it's true that the Oracle case for now is just a US case and in that respect different. But if the Commission had seen a difference between those major jurisdictions, it would have pointed it out. Otherwise all of its statements referred to what was defined as the relevant geographic market in the decision.

2) The Commission's view of the implicit grant was that there was doubt about forks. While the Commission's decision made reference to the scenario of patents being infringed only by newly added code, that was just an example of where the implicit grant would face limits; they didn't say this was the *only* concern. They provided an example on which pretty much everyone can agree, just like you did. But the problem of code changes relating to existing features is a serious one I believe.