[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Proposal to change backport/forwardport procedure
- Subject: Re: Proposal to change backport/forwardport procedure
- From: Chris Travers <..hidden..>
- Date: Sun, 23 Dec 2012 15:49:08 -0800
On Sun, Dec 23, 2012 at 6:59 AM, Håvard Sørli <..hidden..>
I do not think we would (with the current developer base) support more than 2 active releases in the long run. We my have 3 releases for a short term period, during a new 1.x point release.
On 23. des. 2012 13:52, Chris Travers wrote:
support 1.3, 1.4, 1.5, 1.6, and 1.7 at the same time.
In the past we have talked about wanting to promise five years of support for every major version. There are a couple reasons for this. The first is that people may have integration components that they may not want to redo every two or three years. Our current non-destructive upgrade is very good at stock installs but it does require redoing integration components. At one year to release that means about 4-5 releases to support. I think this is also an approach that is likely to get us more developers.
At the current "speed" it takes about a year to release a 1.x point release. It takes 3-6 mounts to gain trust for the new release in the community. A user can choose to do a point release upgrade each year or wait 2. years, and still get support. We my chose to backport fixable security fixes that are part of releases included in Debian stable for a longer period.
I think there is also something to keep in mind. I would expect two things to happen. The first is that bug reports would come in slower over time. Right now we are working through a very few user-contributed bug reports, plus a larger number of bug reports that came from Erik's documentation efforts. At least one of the user-reported bugs is behavior that we inherited from SQL-Ledger. So this is already happening despite the fact that we are still releasing bugfix releases frequently.
So my general thinking is that older branches will at some point (presumably after 2-3 years on average) achieve maturity and at that point I have trouble imagining much in the way of work necessary unless there is a sudden, new security issue discovered that goes way back.