On Sun, Nov 3, 2024 at 9:44 PM Darshan Phaldesai <dev.darshanphalde...@gmail.com> wrote: > > On 03/11/24 7:18 pm, Neal Gompa wrote: > > > On Sun, Nov 3, 2024 at 6:23 PM Darshan Phaldesai > > <dev.darshanphalde...@gmail.com> wrote: > >> On 03/11/24 3:06 pm, Neal Gompa wrote: > >> > >>> On Sun, Nov 3, 2024 at 4:30 PM Darshan Phaldesai > >>> <dev.darshanphalde...@gmail.com> wrote: > >>>> On 31/10/24 5:47 am, Sune Vuorela wrote: > >>>> > >>>>> On 2024-10-30, Darshan Phaldesai <dev.darshanphalde...@gmail.com> wrote: > >>>>>> Few considerations: > >>>>>> First, I plan to include bridges for most libraries needed for > >>>>>> application development and so need to consider their licenses as well. > >>>>>> Second, This project will only hold the "bridge" code and thus users > >>>>>> will still need to install the libraries. I don't think this should > >>>>>> cause any license violations but the bridge code is based of the method > >>>>>> signatures of the original libraries. > >>>>>> Third, Upstream `cxx-qt` project uses Apache-2.0+MIT and I planned to > >>>>>> do > >>>>>> the same but KDE's Licensing policy doesn't mention any thing about > >>>>>> Apache-2.0. > >>>>> I'd say just stick the same license on it as the KDE Frameworks in > >>>>> question. > >>>>> Given they are a directly derived work and also uses the KDE Frameworks > >>>>> underneath, giving any other license is just going to be confusing for > >>>>> the people involved. > >>>>> > >>>>> The app developers needs to deal with (l)gpl licenses anyway. > >>>>> > >>>>> /Sune > >>>> Now that I think about it, this might the best way to maintain license > >>>> compatibility. Individual bridges can be licensed depending on their > >>>> counterparts. > >>>> > >>>> I will make this change. Thank you. > >>> It's not that simple. The reason I suggested MPL-2.0 is because > >>> LGPL-2.1-or-later is effectively the same as GPL-2.0-or-later with > >>> Rust because it's statically linked. > >>> > >>> MPL-2.0 preserves the copyleft at a per source file level, but allows > >>> the binary artifact to have a composition of compatible licenses > >>> (including the GNU ones). This is the least messy for Rust bindings to > >>> LGPL libraries. > >> While most rust executables are statically linked, I am not sure whether > >> this is the case when using `cxx-qt`. This is what running file command > >> on `kontrast-rs` (a sample rust app that uses cxx-kde-frameworks) produces > >> > >> > kontrast-rs: ELF 64-bit LSB pie executable, x86-64, version 1 (SYSV), > >> dynamically linked, interpreter /lib64/ld-linux-x86-64.so.2, for > >> GNU/Linux 4.4.0, BuildID[xxHash]=9e033617833656c9, not stripped. > >> > >> `ldd` also confirms this. I might be totally wrong as I know very less > >> about this. > >> If the final binaries are dynamically linked, the restrictions that you > >> mention still apply ? > >> > > It does, because the *Rust parts* are statically linked, but the > > *C/C++ parts* are dynamically linked. > I see. I did some research online (found your thread as well :) ) and > MPL2.0 makes sense to me. > > Are most framework libraries compatible with MPL2.0? the ones included > at the moment are LGPL2 and shouldn't have any problems. > If no one has issues with it, I will switch to it. >
Yes, it has a high degree of compatibility due to how it works as a license. -- 真実はいつも一つ！/ Always, there's only one truth!
