El jueves, 28 de abril de 2016, 16:09:54 (UTC-4), Mosè Giordano escribió: > > Hi David, > > 2016-04-28 14:10 GMT+02:00 David P. Sanders <[email protected] > <javascript:>>: > > As far as I could see, the original library has an Apache license, so > you should be able to use MIT. > > > > I believe that you need to include a copy of the original license in > your package? > > Original code is dual-licensed, in that case the author of derived > work can choose one of the two licenses or both.[1] I went for LGPL, > that is weakly protective
What is being protected, your own work? but perfectly compatible with MIT "Expat" > No, it is only half compatible with MIT, i.e. only in one direction: you can take code from my MIT-licensed package (e.g. ValidatedNumerics) and use it in yours, but I *cannot* take a piece of code from your package, modify it, and re-use it in mine, without "infecting" my package with the LGPL code, which will automatically mean that I will have to change my license to LGPL. (This is all only my understanding. Please do correct me if I am wrong.) > license, which is the most common in Julia ecosystem, so that most of > Julia package can use this package. Again, they can only *use* it. I cannot even *look* at your code (which I would very much like to do), since if I find something useful, I can't reuse it in my (MIT) package. > For the time being I don't plan > to change the license. > I would urge you to reconsider that. You could, for example, license it with the same dual license as in the original package -- otherwise, you are actually restricting your users more than the original package does (for the reasons I discussed above). The end result otherwise will be that someone who needs to use the functionality will end up going back to the original Fortran code and rewriting what you have already done (which I certainly do not want to do), or using less-good code from somewhere else. > > > +1 for ComplexRoots.jl Steven makes a good point that PolynomialRoots.jl would be a better name.
