On Tue, Jan 21, 2014 at 6:17 AM, Marco Leise <Marco.Leise@gmx.de> wrote:

Am 20.01.2014, 05:32 Uhr, schrieb Martin Nowak <code@dawg.eu>:

Interesting talk by Mozilla's director of release engineering.
http://youtu.be/7j0NDGJVROI

I didn't watch it, because after a few minutes into the talk I realized: A guy from Mozilla talks at Google about the new release scheme they adapted from Google's own Chrome browser and ruined their browser with heaps of regression bugs and confusion about versions. Whatever the success of Firefox is, it was there before they changed their release process. That said it was a lesson learned, so maybe the talk wasn't that bad objectively.

Have the regressions been continuous? Or just during the initial change in process? I didn't notice any stability changes. The version numbers have nothing to do with the talk.

Here is where they show their workflow (have to go back for where they start talking about workflow)
http://youtu.be/7j0NDGJVROI?t=50m24s

It looks like the structure Debian uses

ash => experimental

inbound => unstable

central => testing

aurora => testing

beta => stable

release => stable

The arrows suggest that changes are allows made to the stablest development branch and push back from there, so no cherry-picking.

--
Jesse Phillips