Re: Let's do SOVERSION according to traditions and documentation

On Montag, 23. Februar 2026 13:36:43 Mitteleurop=C3=A4ische Normalzeit Sune=
 Vuorela=20
wrote:
> On 2026-02-23, Vlad Zahorodnii <[email protected]> wrote:
> > In plasma, we operate under different conditions. We can break ABI
> > there. On the other hand, if you consider KF, you cannot break API or
> > ABI compatibility, so you could lock the SOVERSION at the major version
> > and have them both match.
> >=20
> > In plasma, since we can break ABI (we even moved some projects to plasma
> > in order to be able to make breaking changes), locking the SOVERSION
> > like that is simply not applicable.
>=20
> I'm not talking about not doing abi breaking, but doing it in a way
> where you don't make reviewers of the binary packaging stumble.
> We should make it straight and easy.
>=20
> the VERSION and SOVERSION attributes of libraries are not userfacing
> components and we should not treat any of them as such.
>=20
> While the tool allows you to do weird things, one shouldn't.

I agree.
I don't see a reason why to make it confusing, if there is also a simple wa=
y:=20
=2D major version =3D SO version =3D ABI compatible
=2D minor version =3D ABI compatible, may have additions
=2D patch version =3D only minor changes

Alex