Mark Seaborn writes: > David Chan writes: > > Somebody somewhere with a copy of Sibelius will probably save the file > > back into Mudela and stick it up on the web. > > But you could equally well say say that someone [...] will probably > scan the music in and make the Mudela output public. In fact, once > there exists free music recognition software, this is more likely Well ok, I'm sure a free Sibelius import filter is easier to write than a free music OCR program! > > [You could make a secret language compiler to stop people using your > > program source code] > > I'm not sure you can. The GPL says: > ``For an executable work, complete source code means all the source code > for all modules it contains, plus any associated interface definition > files, plus the scripts used to control the compilation and installation > of the executable'' > > [...] I would contend that the phrase ["scripts"] *does* cover the > compiler. I think this is impractical. It would mean that, for example, Windows ports of GPLed software can't make use of proprietory MS Visual C++ extensions. I doubt that's true even for, say, GDK. > > OCR scans are not infallible yet, and it would take quite a lot of > > effort to obtain a [good] source file > > But this is just an incidental problem, not an inherent problem. Some > day OMR will be good enough that it gives hardly any errors. I'm not so sure. You could make a bad-quality smudged photocopy which humans could read but which would be *very* difficult for a machine to read. You can do this with OCR quite easily. You'd need something that would pass the Turing test to be able to read it. With that much intelligence you can look at an assembled binary file and disassemble it into a C program with helpful variable names, etc.. However, I think this discussing the wrong point; see my next comment. > A free music license will have to last, and not contain outdated > restrictions. If OMR gets good enough that it is trivial to read the printed music into a computer, then the printed version *is* "machine-readable source code" and so the restriction doesn't exist. > Say a teacher photocopies a load of free music for her class. If the > music has a distribute-the-source condition, and she doesn't have a > computer, she can't do this for fear that the students will become > awkward and demand the source. Not if the license works like section 3c of the GPL. The original publisher is a commercial distributor and so must, say, put a notice at the bottom of the music saying "download the source from http://funkyedition.com". The teacher is a noncommercial distributor and so only needs to pass on the publisher's offer - which she is doing by photocopying the music. > Why should commercial copiers and non-commercial copiers be treated > differently? Why should commercial copiers be required to have > computers? I don't think this is much of a restriction in today's world! But if you think it is, then we could say that you only have to offer to produce the source if you're distributing a modification (and not if it's in manuscript form). > You become a commercial copier even if you sell photocopied sheets for > a few pence each. I don't think charging people for the medium is commercial copying. (If I charge my friend 20p for a floppy disc with lilypond on it, that doesn't make me a commercial copier). If you're charging much more than the cost of the prints then it is commercial copying. > What incentive is there for a creator of a derivative work to not > release source when he knows that all it takes is one person to make > that source freely-available? Well, look at Mutopia now. There's nothing stopping people from typesetting an old edition of Beethoven's 5th. There is a large demand for copies of this music from orchestras. But it hasn't happened - because it's a huge effort to type all that music in, and not much less effort to scan it using OMR. Believe me, if there were an OMR program available today that made it easy, I would have already done it. The same will happen in future for substantial modifications of our work. Suppose Gershwin had released his Rhapsody in Blue for two pianos on Mutopia, and Ferde Grofe had created his orchestral version as a derivative. Suppose Grofe sold his printed version in the shops - but didn't release the source. How long do you think it would be before somebody typed / OMRed all of that huge orchestral score? It could well be a very long time, if ever. This is hypothetical, but if Mutopia is successful then exactly this situation might arise. If we required the source to be made available, it would make a big difference to how useful modified works like this would be. -- David
