On Fri, Jan 5, 2018 at 2:13 PM, Steve Ebersole <steve(a)hibernate.org> wrote:
While I understand the sentiment of continuing to develop old lines
(branches) of code, that's just not viable. And this is something we have
all discussed as a (full) team a few times now. Once the next development
line is stable we stop developing the older line. That's even more true
within a release family - there is no need to continue to develop 5.x once
5.x+1 is stable. We do adjust that procedure slightly around major
releases, meaning that even after 6.0 is stable we continue to do those
5.x+1 releases *for a short time*
Well, from my experience, it takes some time to get a 5.x version stable.
The first .0.Finals always have a few annoying regressions (not judging,
just stating the fact - same for HV or Search).
And the end users usually wait for the integrators to do the work. You
can't migrate to 5.3 to get your bug fixed if Spring isn't compatible with
it yet (and they usually do the work very early).
I'm not saying we should maintain the versions indefinitely. I'm just
saying that first ".0.Final" != usable.
That's why I'm using the term "consumable by the end users".
Note that in the case of 5.2.13.Final, we don't even have a 5.3.0.Final
yet, so I'm not sure why we are even considering not releasing it.
The last one is from October 19th, we have 50+ issues fixed.
And even the Spring guys are playing it nice asking when we plan to release
it so that they can integrate it in their next Spring Boot release.
--
Guillaume