Re: Introducing ABI changes in libolm

2021-09-17 Thread Daniel P . Berrangé
On Fri, Sep 17, 2021 at 02:52:59PM +0200, Vitaly Zaitsev via devel wrote: > On 17/09/2021 11:48, Florian Weimer wrote: > > Is upstream wrong? Otherwise there shouldn't be anything to take care > > of? > > I think that even minor changes in ABI need a soversion bump. > > Upstream developers decid

Re: Introducing ABI changes in libolm

2021-09-17 Thread Vitaly Zaitsev via devel
On 17/09/2021 11:48, Florian Weimer wrote: Is upstream wrong? Otherwise there shouldn't be anything to take care of? I think that even minor changes in ABI need a soversion bump. Upstream developers decided to hide internal symbols from ELF/DLL export tables. I agree that these symbols are p

Re: Introducing ABI changes in libolm

2021-09-17 Thread Florian Weimer
* Vitaly Zaitsev via devel: > Hello. > > libolm 3.2.6 introduces ABI changes without a SOVERSION bump. > > Upstream's answer: > >> The purpose of bumping the SOVERSION is so that things that are >> compiled against it don't suddenly break. If we're removing things >> that nobody is using (and nobo

Introducing ABI changes in libolm

2021-09-17 Thread Vitaly Zaitsev via devel
Hello. libolm 3.2.6 introduces ABI changes without a SOVERSION bump. Upstream's answer: The purpose of bumping the SOVERSION is so that things that are compiled against it don't suddenly break. If we're removing things that nobody is using (and nobody should ever have used, and people would