Re: Marble versioning
Johannes Zarl-Zierl <[email protected]> Wed, 28 Jul 2021 00:07:29 +0200
| Newsgroups | gmane.comp.kde.kimdaba |
|---|---|
| Message-ID | <542476406.ZSs0jgxMRr@jo2021> |
Am Dienstag, 27. Juli 2021, 15:17:12 CEST schrieb Robert Krawitz: > On 7/27/21 3:08 AM, Tobias Leupold wrote: > > But is it correct that they use the KDE Applications version for the > > library? because if so, the versioning can't be used in a meaningful > > manner, as the "major" version increases each year, no matter if they > > changed something or not, and cmake will declare it as incompatible. > I didn't trace down exactly where it got set, but it does appear to be the > KDE apps version. The versioning is done by the KDE Gear tooling: https://community.kde.org/Guidelines_and_HOWTOs/Application_Versioning > > I think this is simply not a suitable approach for a library ... the > > version should make us able to detect incompatible API changes, whereas > > the version we have now only allows us to see in which year the release > > was done ... this is simply essentially useless for a library, no? > That's definitely (IMO) not how a library should be versioned, for exactly > that reason. Semantic versioning is clearly a better approach if done correctly. If not done correctly (or at all), automatic YYYY.mm versioning is at least reliable and allows us to write checks. Hey, it's not worse than KDE frameworks ;-) Cheers, Johannes