[qalmakka]

You can't just take an MIT/X11 licence, remove it, and relicense it as it's an explicit violation of the MIT/X11 licence. This however only applies to the original code, so any new addition can be under whatever licence you want, including a more restrictive one. That doesn't change your obligation to reproduce the previous copyright notice as instructed by the licence you got the code under

[tannhaeuser]

I get what Nick is trying to do (allow F/OSS to continue receiving security fixes while requiring commercial users to pay), even forwarded his call for help or other support last year. I'm not sure though relicensing MIT code under AGPL is legally sound if your additions are just bug fixes.

[qalmakka]

> I'm not sure though relicensing MIT code under AGPL is legally sound if your additions are just bug fixes.

It is: the original unmodified code is still MIT, but the bits you wrote and thus you hold the copyright for are AGPL. The combination of your bits + old bits is AGPL; some judge may argue your derived work isn't derived enough if you just change a few characters though. Still, I may argue that fixing a bug, no matter how trivial, requires you to perform a creative task, which almost definitely is by itself to be considered enough to go above the threshold required by a derived work.

[eqvinox]

It really doesn't matter until one tries to enforce an AGPL provision. If you do, that's where as you say will be arguments whether one added something substantial enough; basically the AGPL terms can really only be enforced on those additions.

[migueldeicaza]

Correct, not many folks understand this.

[LunaSea]

In practice, how would you differentiate old code from new code when changes start to flow in?

[0x457]

Everything after <GIT SHA> is AGPL unless stated otherwise.