Question about quick release cycle



  • Hi,

    i've been noticing that since 2.1/2.2 there have been relatively many quick point releases.
    also because of the quick releases i notice when the release-date shortens, lots of open bugs get pushed back to a future release.

    i wonder whats better: quickly release without fixing all open bugs or slow releases but have everything thats discovered, fixed before release?

    what do you folks think? personally i don't think i would mind a slower release cycles as lots of specific bugs can be easily patched with system->patches



  • I will vote for "quick" releases - like 6 weeks to 2 months typically - as long as the resources needed to support the formal release build and test process are available. There seem to be reasonably important things that get fixed on that frequency, and actually quite often they are patches to compiled stuff in IPsec, OpenSSL…
    Obviously releases are also needed for FreeBSD security patches that are relevant.
    Some sites are willing to use system patches, but others want to use just official releases - so for those it is convenient to have official releases come out not too long after fixes are confirmed resolved in development testing.

    Of course the other side is that it can be difficult for large production sites to be upgrading to the latest version often - they need to do their own confirmation testing internally... and it takes resources.


  • Banned



  • If you wait until every known bug is fixed to release, no one would ever release any software (unless they're just ignorant of what bugs exist). Perfect is the enemy of good in release engineering - no one ever has, or ever will release completely perfect software.

    2.2.3 is currently better than 2.2.2 in every way we're aware of. There are important fixes that need to get out to people. So that's where we've drawn the line with getting release out. Some bugs always have, and always will, get pushed to some future release.


  • LAYER 8 Netgate

    I am just hoping that if anything significant happens security-wise we will see an update for 2.1.5.  Until limiters are fixed 2.2 is simply not an option (at least for me in a couple cases.)



  • ok thanks all, for your input in this matter.


  • Rebel Alliance Developer Netgate

    @Derelict:

    I am just hoping that if anything significant happens security-wise we will see an update for 2.1.5.  Until limiters are fixed 2.2 is simply not an option (at least for me in a couple cases.)

    Highly unlikely. We were hoping 2.2.3 would be the last 2.2.x release even, so we could focus more on 2.3. But 2.2.3 has a couple small issues that need addressing so we'll be putting out a 2.2.4 in the near future.

    I'm all for releasing more often. In addition to keeping everyone more up-to-date, so long as there are few if any regressions then it's better than letting bugs or security issues sit out there for longer periods of time.

    Some people also fall into the false trap of thinking a project is inactive if they do not release very often. If you check our repositories you will see we are very active, but when we go for long periods between releases people seem to think the project is dead or suffering, which is not the case.


Log in to reply