Hey Balázs,
Well the current LTS positioning is (as mentioned in that info note on downloads page):
- Released every 18 months, Long-Term Supported versions are the most stable releases to which all critical and blocker issues are back-ported.
Meaning that indeed the LTS is often considered as the version of choice if production stability is a prime concern. To share some personal thoughts though, you could read stability in two ways: technical stability (the less bugs the better) and UX/functional stability (UI and feature offering not changing too quickly). And this is where I believe SonarQube is evolving as it grows up:
-
looking back years like 2015/2016, SonarQube has built-up lots of confidence on its core offering (bug/vuln/code_smell detection, great methodology with the leak period etc.) and you can see the UI/UX went through a number of evolutions to support that, and make it an even stronger tool
-
some of those changes couldn’t happen overnight, and it did happen that a ‘vision’ (in terms of product evolution) would take a few intermediary SonarQube versions to materialize, with the goal of then delivering a solid new LTS
So back then you indeed wanted to consider those intermediate versions with quite some care (and ourselves were advocating our customers to be careful about their choices). Now if we look at the current landscape:
-
SonarQube UX has solid foundations which are in place for a while now, with functional improvements smoothly built on top of it (take native branch support for example)
-
each intermediary version has a solid value proposition (see latest announcement for example) and should by no means be considered half-baked. We pay very close attention to that, not just for SonarQube actually, but simply because we’re continuous delivering on SonarCloud, so it only makes sense that we then open up these cool new feature on SonarQube.
So if I were to echo back my initial point, I’d say that technical stability still is a de-facto attribute of an LTS (by nature, given the critical bug-fix backport model). However the UX/functional stability is arguably less a reality nowadays, and even if a fraction of it remains, not sure it prevails when balanced against the quantity and rate of cool new features delivered in latest versions.
I realize this is a bit of a ‘longer than usual’ response, but saw it as a nice opportunity to give some deeper insights into how SonarQube is growing.