Is the code generated from a GPLv3 EBNF grammar a derivative work?

Question

So, I'm really unclear on this situation and what it means.

I have an EBNF-based grammar file that we use for our project. This EBNF grammar describes rules for parsing/lexing a document (e.g. some source code for a Domain-Specific language). EBNF can be thought of as a generalisation of regular expressions.

Lexer/parser generators, such as ANTLR, YACC, and Bison, take this grammar file as input, and output source code which can perform the parsing of the language described by the grammar. We are using ANTLR, it is outputting a suite of C# classes.

The particular grammar file we use as input is licensed under GPLv3 and we've made modifications to it.

Now, the resulting files do not require the grammar file after the classes are generated. Technically, there's no need to distribute the grammar file along with the project. We could keep that private and distribute everything without it and everything works just fine.

Up until now, we've been releasing our software under GPLv3 under the assumption that our project is a derivative of this grammar file, but is it? I'm wondering if we're really required to release our entire project under GPLv3. We'd much prefer to release under the MIT license. Can we do that? The grammar file isn't a library, but a tool that is used in conjunction with ANTLR to generate code.

Can we create a separate distribution/repository for the modified grammar file, then re-license our core project? (We'd still like to share our modifications to the grammar, we just don't want to continue licensing ours with a viral license).

Answer 1

The code is a derived work of the grammar. (but not of ANTLR)

• The Grammar file is a description of how to parse the language, written in EBNF.
• The source code generated by ANTLR etc is a description of how to parse the language, written in Java (etc).
• The source code was derived from the Grammar, using ANTLR.
• It is thus a derived work, of the Grammar (but not of ANTLR).

It would make very poor sense for ANTLR to try to claim that anything made using it was a derived work. Like painting is not a derived work from a paint brush, or a letter a derived work of the word processor. I won't go into that further, (I think that further details should go in another similar question with different emphasis)

The GPL uses the term "work based on" rather the derivative work, but I believe though are synonymous Thus derivative works must be licensed.

---

Consider translating between two main stream programming languages. (EBNF is essentially a programming language, as is Java.) The translated source code is clearly a derived from the original.

To give a concrete example:

• I have some source code, written in Dart I used dart2js to convert it to Javascript for execution in the browser.
  • My Dart source code, is analogous to your Grammar file
  • dart2js is analogous to ANTLR -- changing one language to another
  • the Javascript output which I incorporate into my overall webapp is the Source Code, which you incorperate into your project
  • If I released the Dart code under the GPL, and then someone else run dart2js on it and said it was a original work; well they would be incorrect, and obviously so. (NB: I have not released the linked code under GPL or any other license)
  • The source code is not a derived work of

---

Consider also the non-software examples:

• I have written a play. I producted a Play script in English, and it was translated to Russian.

  • The play script, is a set of instruction on how to produce the play. How to move on stage, what to say when to say it etc.
  • The English playscript is your Grammar File.
  • The Translator is your ANTLR
  • The Russian playscript is your Source Code produced.
  • The Russian playscript is a derived work from my English original.

• Or I have a music score (sheet music) it describes a song in musical notation.

  • I have my band play the song, and I record it into a MP3.
    • I now have a description of the song, in MP3 form
    • That recording is a derived work
  • Or, say I used OCR tool to convert it to MusicXML
    • The sheet music (score) is exactly analogous to your EBNF grammar
    • The OCR tool is analogous to ANTLR
    • The MusicXML file generated is analogous to your generated class files
    • The MusicXML was derived from the score

---

Taking one work, and using it as the bases for another work is creating a derived work. It doesn't matter if it looks different, it is still derived.

Your modified grammar would still have to be licensed under the GPL if released (it is obviously derived, as given by the words "modified"), as would any code generated from it, thus your whole project would have to be under the GPL.

The short of it is: The Grammar is under a viral license, therefore anything derived from it is under that license (otherwise what would be the point in licensing it under a license that does not allow relicensing?). Producing source code is producing a work based on the grammar (like any translation). Thus the source must be under the viral licence

(But, I'm not a lawyer and I am not your lawyer)

Answer 2

TL;DR: yes, the software you distribute which builds on grammar files under copyright for which you have a GPLv3 license must also be licensed under the GPLv3.

Your software is made up of several parts.

• The ANTLR runtime (BSD)
• The original grammar files (GPLv3)
• The rest of your software

Your software contains in part a machine translation of the grammar files. The US copyright office has said

in order to be entitled to copyright registration, a work must be the product of human authorship. Works produced by mechanical processes or random selection without any contribution by a human author are not registrable.

meaning that the generated parser is not a separate derivative work as far as copyright is concerned, it is the same work.

ANTLR also injects large parts of its own codebase in the generated parser, and links against the ANTLR runtime. This is not much of a problem as both are under the (permissive) 3 clause BSD license, which is easy to comply with in combined or derivative works.

What I'm not sure of, and what I believe could only be tested in court, is whether grammar files themselves would be eligible for copyright protection at all. What is copyrighted is the expression of the idea the grammar describes, but one could argue there is no originality in the expression of the grammar. This would probably also depend on the grammar file itself. It might make a lot of difference if it's just BNFR rules or if there are significant code snippets or other things that are definitely eligible for copyright in the grammar files.

What you could certainly do is make a clean room re-implementation of the grammar files, and you're not bound by the license of the grammar files.

Answer 3

I dug into this and found some answers in the GPL FAQ.

Is there some way that I can GPL the output people get from use of my program?

For example, if my program is used to develop hardware designs, can I require that these designs must be free? (#GPLOutput) In general this is legally impossible; copyright law does not give you any say in the use of the output people make from their data using your program. If the user uses your program to enter or convert his own data, the copyright on the output belongs to him, not you. More generally, when a program translates its input into some other form, the copyright status of the output inherits that of the input it was generated from.

So the only way you have a say in the use of the output is if substantial parts of the output are copied (more or less) from text in your program. For instance, part of the output of Bison (see above) would be covered by the GNU GPL, if we had not made an exception in this specific case.

You could artificially make a program copy certain text into its output even if there is no technical reason to do so. But if that copied text serves no practical purpose, the user could simply delete that text from the output and use only the rest. Then he would not have to obey the conditions on redistribution of the copied text.

http://www.gnu.org/licenses/gpl-faq.en.html#GPLOutput

Can I use GPL-covered editors such as GNU Emacs to develop non-free programs?

Can I use GPL-covered tools such as GCC to compile them? (#CanIUseGPLToolsForNF) Yes, because the copyright on the editors and tools does not cover the code you write. Using them does not place any restrictions, legally, on the license you use for your code.

Some programs copy parts of themselves into the output for technical reasons—for example, Bison copies a standard parser program into its output file. In such cases, the copied text in the output is covered by the same license that covers it in the source code. Meanwhile, the part of the output which is derived from the program's input inherits the copyright status of the input.

As it happens, Bison can also be used to develop non-free programs. This is because we decided to explicitly permit the use of the Bison standard parser program in Bison output files without restriction. We made the decision because there were other tools comparable to Bison which already permitted use for non-free programs.

http://www.gnu.org/licenses/gpl-faq.en.html#CanIUseGPLToolsForNF

Emphasis mine, as it's the important part of this FAQ. Because Antlr copies large parts of the grammar file into the resulting parser classes, those resulting classes are licensed under whatever license the grammar file uses. I find it interesting that another such tool (Bison) was mentioned here and given an explicit exception. I could find no evidence of any such exception for Antlr, so any code that links to the generated parser, must be licensed as GPLv3 like the grammar file.