[llvm-dev] RFC #2: Improving license & patent issues in the LLVM community

Daniel Berlin via llvm-dev llvm-dev at lists.llvm.org
Thu Nov 3 10:11:51 PDT 2016

On Thu, Nov 3, 2016 at 9:13 AM, David Chisnall <David.Chisnall at cl.cam.ac.uk>

> On 3 Nov 2016, at 15:55, Daniel Berlin <dberlin at dberlin.org> wrote:
> >
> >> I and the other libc++ contributors were all happy to have our code
> relicensed under the MIT license (or contribute it under those terms
> originally), so there is some pretty clear evidence that we explicitly did
> *not* require this attribution.
> >
> > That license does, in fact, require attribution:
> > "The above copyright notice and this permission notice shall be included
> in all copies or substantial portions of the Software. “
> Again, unless I’m misreading this, there is no requirement that code that
> merely uses libc++ headers include a libc++ attribution.

You are misreading this :)

I can tell you it's not even a consensus, i have not found a single lawyer
who believes otherwise.

I can forward you the discussion that occurred around this if you like (i
have to acquire permission from the folks involved first). There was 100%
universal agreement on this.

> That was the explicit requirement for the switch to the MIT license.  The
> proposed new license would mean that this is only true for code compiled
> with LLVM.
> We must include the copyright and attribution when we distribute FreeBSD,
> because we include the whole of libc++.  We do not need to separately
> include it in programs that simply link to libc++ (irrespective of how they
> were compiled) in packages, because they do not contain ‘substantial
> portions’ of the code (probably - it’s a bit of a grey area with respect to
> template-heavy C++).

This is not correct ;)

> >> Indeed, we relicensed under the MIT license (after long discussion)
> specifically to avoid this requirement.
> >
> > If we did, we screwed it up then. We probably meant the zlib license,
> which does not require attribution  ;)
> Assuming that your interpretation of this clause - which is the first time
> I have ever heard it interpreted to include either linking or including
> headers - is correct.

This is interesting, as i've said, i've literally never heard it
interpreted otherwise.

> I’d be quite surprised if it were, as the lawyers that reviewed it at the
> time were happy that it did not impose these constraints, but it’s possible.

Like I said, there was discussion of it on the open source counsel mailing
lists i belong to, and there was universal agreement that it required
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20161103/fb0c9cd9/attachment.html>

More information about the llvm-dev mailing list