Attribution when deriving from a W3C Software Notice and Document License work

Question

I am creating a reimplementation for a different language of a software library that is licensed under the W3C Software Notice and Document License. The original library contains a NOTICE file. As my project is heavily influenced by the original I would assume that it constitutes a derived work (which I want to publish under the MIT license).

I therefore wanted to ask for some advice on how to properly attribute the original library. Is it enough to include the text This software or document includes material copied from or derived from [title and URI of the W3C document]. Copyright © [YEAR] W3C® (MIT, ERCIM, Keio, Beihang). from the license in my README and the license text in the LICENSE file? Or do I have to create a NOTICE file myself (containing the text from the original NOTICE) and/or add the above mentioned copyright statement to every source code file in my library?

Answer 1

The W3C license you linked to does not mention a NOTICE file, so there is no requirement to have or keep such a file.

The best way to attribute the original library is to have the text This software includes material copied from <original library> (<link to repository>). Copyright © [YEAR] W3C® (MIT, ERCIM, Keio, Beihang). in both your README and LICENSE files.

You put it in your README to have the attribution in a location where it will be almost certainly seen.

You also put it in your LICENSE file, just before the copy of the W3C Software Notice and Document License itself to make it clear why that license text is placed there and to avoid any possible confusion if your code might be dual-licensed or not.

Answer 2

I am creating a reimplementation for a different language of a software library that is licensed under the W3C Software Notice and Document License. The original library contains a NOTICE file. As my project is heavily influenced by the original I would assume that it constitutes a derived work (which I want to publish under the MIT license).

Legally, I am not sure it is a derived work. Ask your lawyer. But for ethical reasons, it probably is. A counter example is obvious: many C compilers (like GCC, Clang, tinycc, etc...). They all are doing the "same thing" but are legally not a derived work (the same ideas are implemented in several of them). Another counter example are Unix shells (like GNU bash, dash, zsh...). A lot of them are able to run the same scripts (POSIX shells).

Another example is musl-libc and GNU libc. They implement the same API. They have different licenses and copyright owners and contributors.

Yet another examples are GUI libraries like FLTK, GTK, etc. All of them are able to show windows on screens. And Wt and Qt have very similar APIs. But Wt show quasi windows in a web browser, but Qt is showing windows on a graphic server (e.g. Xorg).

In France (where I work and live) you could become member of APRIL and ask them. They do have legal expertise and political influence in favor of open source.

My recommendation is to add some copyright notice in every source file (like the GPL suggests). Things are more complex for bootstrapped compiled implementations. I usually generate a GPL comment in generated C or C++ code.