Jesus Cea <jcea%jcea.es@localhost> writes: > Thomas, I see this change 15 minutes ago > <http://cvsweb.netbsd.org/bsdweb.cgi/pkgsrc/databases/py-bsddb3/Makefile>: > > """ > Allow db6 and set up the environment correctly so this builds > automatically against it. > > Users have to have accepted the db6 license already for installing it > anyway. > """ > > That is not right. > > If you use db6 and you don't pay Oracle, your program *MUST* be AGPL3. > The reason to require a license acknowledge is for the programmer to be > aware of this imposition. > > Note also that if a program I wrote works uses db4 and you "upgrade" me > under my feet to db6, suddenly my program is breaking Oracle licensing > unless my code was AGPL3 licensed already. Two nits (and IANAL, TINLA, etc.) There is no "must" in copyright law; there is only "may not". But agreed that it amounts to close to the same thing. AGPL3 terms are only triggered by redistribution or making available over a network. But other than that, you are 99% correct that this puts users in a bad situation. So far, pkgsrc has avoided making decisions about whether DEPENDing or buildlinking with a package creates a derived work that has a different license than the original terms. I don't want to go down that path. The real issue here is AGPL3 use aimed at relicensing, rather than AGPL3 as intended for web applications themselves. I sent a note about splitting the license tag. This situation argues for that (and for people to decline to use db6 or other libraries that are AGPL3/relicensable).
Description: PGP signature