The docs on the Boost license say as much as well, and derive from legal consult. I must say that after reading this I felt a lot better about the headers I've implemented.
Sent from my iPhone On Jul 9, 2011, at 1:56 AM, Mike Parker <aldac...@gmail.com> wrote: > On 7/9/2011 5:43 AM, Steven Schveighoffer wrote: >> On Fri, 08 Jul 2011 16:02:39 -0400, Johannes Pfau <s...@example.com> wrote: >> >>> Steven Schveighoffer wrote: >>>> On Fri, 08 Jul 2011 15:39:22 -0400, Johannes Pfau <s...@example.com> >>>> wrote: >>>> >>>>> Andrej Mitrovic wrote: >>>>>> What's the license on the bindings? >>>>> >>>>> Have not thought about that yet, but I think I'll use the boost >>>>> license. (I'm not sure if that's possible, as >>>>> cairo is LGPL, maybe I'll have to release the binding part at least >>>>> as LGPL, as that's based on the cairo headers? Stupid licensing >>>>> stuff... ) >>>>> >>>> >>>> I'm not a lawyer, but I think LGPL just covers the library code, not >>>> the bindings, as long as the link is dynamic. In other words, LGPL >>>> specifically allows dynamically linking with any license, as long as >>>> the library remains LGPL. >>>> >>>> For reference, the C standard library (which phobos uses extensively) >>>> is LGPL on Linux (glibc). >>>> >>>> -Steve >>> >>> True, but in this case the bindings were translated from the cairo c >>> headers. I wasn't sure if the bindings could be considered a derivate >>> work. >> >> Well, if we look at this logically -- if the C headers contain so much >> code that using them would require releasing your software under LGPL, >> then why even use LGPL? The only point for using LGPL is to allow other >> licensed code to use your library, yet still have your library be under >> the GPL. If linking a C application using the C headers doesn't require >> GPL'ing your code (or LGPL'ing), then I can't see how a translation of >> them would require it. >> >> But almost certainly a translation of the headers is a derived work, so >> the bindings themselves should have the same license as the headers >> (LGPL). I think this should cause no problems with linking proprietary >> code. I don't think it would qualify as a phobos module though. >> > > A few years back I was concerned about this same issue with my SDL bindings > in Derelict. SDL was licensed under the LGPL. So I posted a question to the > SDL mailing list. The maintainer, Sam Lantinga, answered thus: > > "The API is not copyrighted, only the SDL implementation is. I would > consider this a work that uses the library, rather than a derivative work. > You'll notice that using inline functions in LGPL headers, which technically > places code from those headers in your object code, also does not change your > work into a derivative work - it remains a work that uses the library." > > That reinforced my own understanding. So for years now the license for > Derelict's bindings has usually been different from the C libraries. That's > especially true for Derelict 2, which is licensed under Boost. > > Realistically, I'd like to release Derelict with no license at all. I'm not > sure exactly what it is I'm licensing. The lion's share of original code in > the project is in the utility package. The bindings themselves generally have > very little non-interface code, just what's required to implement the loader. > But I've learned that people get nervous when there's no license attached to > a library. So Boost it is.