On Fri, 16 Oct 2020 at 09:47, Paul Wouters <[email protected]> wrote: > > On Fri, 16 Oct 2020, Tuomo Soini wrote: > > >> There were thoughts about removing IPSECBASEVERSION alltogether, and > >> setting it now to something like "4.1dev". > > > > Just 4.1 - the next version > > That would be misleading though, you would see 4.1-gXXXXXX, thinking > it is post 4.1 ?
Yes. Same applies to 4.1dev. It minimally suggests there's a 4.1 branch and probably a 4.1 release. Basic problem is that the tag is trying to predict the future. Perhaps the first thing to do is look at the release process itself. Currently it drops a tag. Should it instead cut a short lived branch? How do other projects deal with this? (btw, kvm uses testing/web/gime-git-branch.sh and deals with hairy edge cases because HEAD in the repo-under-test may may be detached) > >> While still merging the git short commit hash into that to know the > >> exact commit of the dev tree the build was on. > > > > When building from git we know we are in snapshot because there is no > > tag for v4.1 > > I see. But then when you build from the spec files, it somehow has to > show that too, and we normally do that by using a rcX or drX pre-release > marker. That is why I was thinking of using the "dev" marker. > > Paul > _______________________________________________ > Swan-dev mailing list > [email protected] > https://lists.libreswan.org/mailman/listinfo/swan-dev _______________________________________________ Swan-dev mailing list [email protected] https://lists.libreswan.org/mailman/listinfo/swan-dev
