> The piece of code in question was dual licensed, both GPL and BSD, which makes it
> fully legit to distribute derivate work under the GPL license.

I wouldn't count on it. Only the copyright *owner* can relicense the code; removing
the BSD license is no more legitimate than removing the GPL from a GPL-only module.

> Of course I can understand Theo de Raadt feelings concerning the issue, but perhaps
> they should have put another license on the code instead. I don't see how this is
> worse than for example creating a derivate work of the whole OpenBSD kernel and
> release it in binary form only and not allowing users the source at all, which the
> BSD license support. If they get so upset about this, perhaps they have to
> reconsider their licensing scheme completely.

It is *worse* because the GPL community claims "high moral ground" on code sharing;
changing the license in order to prevent the derived work from being used on the
same terms the original was offered is contrary in spirit to the ethos which the GPL
is supposed to be upholding.

One nasty side-effect of the "theft" here is that it balkanizes the development
community: those who originated the code, and who continue to maintain it under
its original license, are artificially isolated from the work of those who are
relicensing it.