Hi. Le jeu. 15 déc. 2022 à 17:36, Mark Thomas <ma...@apache.org> a écrit : > > On 14/12/2022 12:12, Gilles Sadowski wrote: > > Hi. > > > > Le mer. 14 déc. 2022 à 12:25, Gary Gregory <garydgreg...@gmail.com> a écrit > > : > >> > >> I would create a branch called "1.x" instead and bump the version in the > >> POM to 1.5.0. > >> > >> FYI, I've been using the x.y.z version format in most of not all components > >> I work on, I find that it sets expectations better, for me anyway. > > > > IIRC, the convention is to use "x.y" if "z" is "0". > > I have a very strong perference for always including the z component > even if it is zero. That said, for consistency I intend to follow > whatever pattern (I haven't looked yet) file upload used for the > previous 1.x releases. > > > If a third number refers to "patch" or "bug fix", and there hasn't > > been any, it is rather more confusing. > > I disagree. I see greater confusion between 1.y referring to the > specific 1.y(.0) release or the series of 1.y releases. If the .z > component is always present, there is no opporutnity for confusion. > > > IMO, this is the kind of thing that should be consistent across all > > releases within a project; so departing from the common (and > > Commons') practice should not occur without a vote. > > If the process wasn't established with a VOTE there there is no need for > a VOTE to deviate from it. > > It seems reasonable that there should be a consistent versioning format > but it isn't clear if we'd be able to reach consensus on what that > should be. > > > Perhaps the same remark about naming (git) "tags". > > Ditto.
My point is: either we collectively follow (informal) convention, or we propose a change, possibly with a vote to make it a formal convention. I don't mind changing the version naming scheme because having a convention is beneficial (meaning: All components should do the same, to avoid confusion). For example, a few years ago, an alternative model was proposed for git development (and used in the Commons Math repository). Notably, in that model, the "master" branch was named "development". That proposal was criticized, not on the merit of the model, but on the sole fact that the (informal) convention was to have a "master" branch. And the experiment stopped, just to maintain consistency with the other components whose contributors did not see it useful to depart from the convention. Gilles --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
