Re: [Haskell-cafe] Re: ANN: hakyll-0.1

On Tue, Dec 8, 2009 at 4:13 PM, Robert Greayer wrote:

On Tue, Dec 8, 2009 at 4:46 PM, Tom Tobin wrote:

...

On Tue, Dec 8, 2009 at 3:30 PM, Ben Franksen wrote:

...

Ketil Malde wrote:

...

Your contributions could still be licensed under a different license (e.g. BSD), as long as the licensing doesn't prevent somebody else to pick it up and relicense it under GPL.

At least, that's how I understand things.

Right. So hakyll is absolutely fine with a BSD3 license, AFAICS.

Seriously, no, this is *totally* wrong reading of the GPL, probably fostered by a misunderstanding of the term "GPL-compatible license". GPL-compatible means the compatibly-licensed work can be incorporated into the GPL'd work (the whole of which is GPL'd), *not the other way around*.  If you are forming a derivative work based on the GPL'd work, and thus you have to release that derivative work under the GPL.

The crux here is that the source code of hakyll, released on hackage, is not a derivative of Pandoc (it contains, as far as I understand it, no Pandoc source code).  A compiled executable *is* a derivative of Pandoc, so anyone who *distributes* a compiled executable would need to make *all* the source available under the GPL (including the hakyll source).  Since the hakyll package is released under BSD3, this would be allowed (AIUI, IANAL).

IANAL either, but my understanding is that judges take a very dim view of attempts like this to evade the requirements of a license. If a piece of software is built on another piece of software, it doesn't matter if you're looking at source code or a binary. I can write the SFLC and pose a hypothetical situation that captures the gist of what we're talking about, and post the response here, if anyone is interested.