Thunderbird betas - proposal: move to week 2 of a cycle

Wayne Mery (d531) vseerror at Lehigh.EDU
Mon Nov 19 12:02:30 UTC 2012


On 11/19/2012 6:35 AM, Mark Banner wrote:
> Hi All,
>
> With the move to the new release model, and as per the plan, we're still
> expecting to do new beta releases once per cycle.
>
> I would like to propose that we move the beta release to the second week
> of a gecko rapid release cycle - i.e. although Firefox would release a
> new gecko beta around the same time as a new release point (e.g. 18.0b1
> normally comes out a day after 17.0), we'd actually wait a week before
> building and releasing Thunderbird 18.0b1.
>
> This idea has been suggested to me a few times, and I'm now thinking it
> would give us some advantages. Here's the list:
>
>   * Any big issues found with Gecko in Firefox's 18.0b1 would likely be
>     fixed or at least known, before we produce the Thunderbird 18.0b1
>   * If a Thunderbird release, e.g. 17.0 or any other, requires a
>     chemspill due to security or stability issues, this may give us
>     flexibility around producing a beta to test fixes for the
>     security/stability issues if we needed to use the beta
>   * Also, wrt chemspills, it would mean that we would not need to
>     produce both a 17.0 and 18.0b1 on the week of release, and then
>     17.0.1 and 18.0b2 the following week (we'd just do 17.0, 17.0.1 and
>     18.0b2)
>   * There would be one less release to track and manage during release week
>   * It gives L10n another week for any sign-offs to get their work into
>     beta (it'll be continuously integrated anyway, but we do see some
>     sign-offs in the first week or so after a merge day)
>
> For Gecko 24, we'd probably move the first beta to the first week again
> (as it will form the release), which will mean a slightly shortened
> Gecko 23 test cycle. However, I don't see a big issue in doing that.
>
> Note that the repository merges would still happen on the same day as
> they do now, i.e. in sync with mozilla-*.
>
> If there's no objections to it, then we'll do betas in the second week
> going forward.
>
> Mark.

I'm all for it, given the potential for increased stability of initial 
release of betas.

It also gives addon tiny bit more time to be updated - if they are able 
to test using the final alpha.  Are there any further considerations 
with regard to addons?



More information about the tb-planning mailing list