Features vs. stability

I can see this tension in multiple forum topics: apparently the current testing branches are unpopular, and many people aren’t really interested in the new features (at the particular moment; often including myself). I think the project has matured to a point where it would be good to consider having some “longterm” release that would only contain security and other critical fixes.

Of course, it would cost the team more time (which they lack), assuming the quality of other releases should remain. Here I personally think the quality of “other releases” would be of less importance, as (1) it would be voluntary thanks to the “longterm” alternative, and (2) people wanting new features tend to have Linux skills above average.

Another way of achieving a similar effect is to slow down the release cycle. Currently it’s around two months IIRC, but it could be half a year (for example, for now), giving some overlap so people wouldn’t be forced to a big upgrade immediately. People could be testing new features on what we currently call “RC” branches – they would understandably roll for longer time and the name probably wouldn’t fit much anymore.

Let’s roughly estimate preferences of the user base:

  • I prefer having a “longterm” branch, in some way (see above), despite splitting the effort of the team.
  • I prefer the current approach to branches.

0 voters

(about) I’m an active contributor to a Linux distribution community, remembering lots of discussions and issues around the stable release branches introduced several years ago. In the upcoming year I’m officially coordinating those releases. EDIT: this post is my personal opinion, and I’m not part of the Turris team proper.


I think a more important question is why the team doesn’t have time to effect reasonable QA.

Omnia was an incredibly successful crowdfunding campaign, yet we still are pushed updates which have not been even basically tested (I run an almost-vanilla Omnia because I want the most-tested configuration- yet today I don’t have a Foris interface).

What happened to all that money? Was too much allocated to fancy features rather than basic stability?

Regarding your poll, I don’t care about release cycles. I just expect a smoke test that a vanilla device untouched by human hands does not lose basic functionality when updated automatically. Having the device Foris UI disappear is utterly unacceptable and cheaply preventable.

Not having a test device running unattended as a guinea pig for new releases is utterly ridiculous and shows contempt for the users who’ve paid a lot of money for this hardware and made the crowdfunding so hugely successful.

I’m running an Omnia in almost basic config myself, and I don’t remember ever experiencing such a blackout… so your problems apparently don’t seem to be as easy to detect by testing as you might expect. As for the money, my personal guesstimate is that a rather large fraction of it got spent just on the expensive Omnia hardware that supporters “bought” in the campaign.

My point in this thread is that typically fewer changes mean fewer opportunities for things to go wrong, and security patches tend to be the less intrusive changes, too. The way QA is done seems a different topic and mostly independent of this thread’s question, so better move that discussion into a separate thread…