On 15/09/10 15:03, Julian Taylor wrote: > On Wed, 2010-09-15 at 15:07 +0200, Mateusz Loskot wrote: >> On 15/09/10 15:04, Denis Arnaud wrote: >>> 2010/9/15 Mateusz Loskot <[email protected] >>> <mailto:[email protected]>> >>> >>> For instance, it could be managed this way: - bug fixes bump only >>> patch level (micro version number), never bumps SOVERSION - >>> changes to existing core and backends that touch API or ABI only >>> bump minor level only, bumps SOVERSION - addition to the SOCI >>> system as a whole like new backend, new data mapping (e.g. high >>> precision artithmetic numbers) would bump major version number, >>> bumps SOVERSION as well. >>> >>> Makes sense at all? >>> >>> >>> I agree for all those three points. >> >> Good. >> >> I'd like to wait for comments from Julian and Artyom and anyone >> else who is interested in this aspect. Let's make it based on a >> democratic decision process, so all who participate in it will be >> happy :-) > > > Having a more complex soversion is not recommend, but also not > forbidden, by debian policy. > > So as long as you will always break api/abi on minor updates and > never on patch releases its fine with me.
Julian, Sorry for later respons. Yes, patch releases must never break API/ABI. However, after learning a bit more about KDE versioning, I've started to appreciate its simplicity. Perhaps it is a good idea to slightly modify my proposal to follow KDE's scheme. I'm not KDE, but GNOME user and I think both desktops use the smae scheme: http://live.gnome.org/InterfaceSpecification#Library_Versioning Applying to SOCI, we would get the following 3 simple rules manually maintained by the team and automatically executed by CMake: MAJOR - Incremented when the interfaces are changed in ways that are ABI incompatible. When incremented MINOR and PATCH become 0. MINOR - Incremented when interfaces are added or deprecated. In other words, only ABI compatible interfaces changes are allowed. When incremented PATCH becomes 0. PATCH - Incremented when only bugs that do not affect interfaces are fixed. Requesting for comments guys! Best regards, -- Mateusz Loskot, http://mateusz.loskot.net Charter Member of OSGeo, http://osgeo.org Member of ACCU, http://accu.org ------------------------------------------------------------------------------ Start uncovering the many advantages of virtual appliances and start using them to simplify application deployment and accelerate your shift to cloud computing. http://p.sf.net/sfu/novell-sfdev2dev _______________________________________________ Soci-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/soci-users
