On Fri, May 30, 2014 at 7:54 AM, Ian Kelly <ian.g.ke...@gmail.com> wrote:
> On Thu, May 29, 2014 at 1:40 AM, Chris Angelico <ros...@gmail.com> wrote:
>> If you absolutely can't get in touch with him, the only option is to
>> go back to the original protocol and manually reimplement it,
>> completely ignoring this code. It's sad but true; some code dies
>> because of a trivial thing like "Oops, I forgot to actually say that
>> this is MIT-licensed".
> The second part of that is that the code should actually *include* the
> license text. Just writing "BSD license" somewhere on the website or
> in package metadata is annoyingly common but somewhat questionable in
> how a judge might interpret it. For instance, there at least four
> different versions of the BSD license; which one did you mean?
Agreed. I tend to have a file called README or LICENSE in the main
source code directory that has license terms (in the case of a README,
the license will follow whatever else there is to say); that's
generally clear enough, without having to put a header on every single
I like to put both a short name and the full license text in there
(see eg https://github.com/Rosuav/Yosemite for which I use the MIT
license), so it's independent of random web sites - having nothing but
a link to the license text makes it that bit more vulnerable.