A question, of sorts.

[Simon Stapleton]

Does it matter if he still violates the GPL? No it does not. When he admitted in public for all to see that he violated the GPL, he lost his rights the GPL license gave him. This does in fact mean every kernel released since is in fact a violation of federal copyright law.

…

This will remain the case until every (not 99.9%, but 100%) copyright holder agrees to grant a new license.

Now, it's my understanding (which is perhaps totally wrong) that:

A person, let's call him/her "P", distributes a binary derivative work of GPLed software, and refuses requests for source.

This places P in violation of the GPL; at this point they have forfeited the right to use or distribute the original software. P, however, has every right to use his/her own changes to that software, let's call it "patch-P", because those changes are copyright P.

So, assuming P was working from original version X.0 of the original package (which he/she has forfeited rights to), does this mean that if he/she downloads version X.0.1 of the original package, *receiving, in the process, a new grant of rights under the GPL*, he/she has no rights to use that either? Eric appears to think that P has forfeited his/her rights under the GPL for version X.0 *and* all or any subsequent versions, forks or derivative works, forever. Perhaps even *any* GPL software. I don't think that's reasonable, and I can't find a reading of the GPL to support it.

In any case, of course, should P then distribute a binary of X.0.1 + patch-P and refuse requests for source, then that is also in violation.

I also can't find any reading that requires the agreement of 100% of copyright holders in the original package before a new license is granted. Based on what's happened with companies "coming back into compliance", all that seems to be required is "fixing the situation" going forwards (not necessarily even rectifying past non-compliance beyond withdrawing non-compliant binary releases).

Am I wrong?

regards

Simon