[Opendnssec-develop] Re: More on OpenDNSSEC versions... Support policy and 1.5 anyone?

[Roland van Rijswijk]

Hi Sara,

Op 3 apr. 2013 om 17:21 heeft Sara Dickinson <sara at sinodun.com> het volgende geschreven:

> In the team meeting last week we had some more detailed discussion about the impact of the change in versioning scheme as recently agreed. 
> 
> 1. Support for older versions
> 
> The consensus was that a significant portion of the user community will stay on 1.3 for the foreseeable future and that we should consider declaring it a Long Term Support release (following the Ubuntu model).  
> 
> - It was proposed that in practice this would mean committing to providing support (i.e. bug fixes only) for 1.3 for at least 1 year after 2.0 is released (assuming 2.0 is the next LTS). 
> - 1.4 and any other 1.X minor releases would be 'standard releases'. 
> - Support for standard releases would be something like N months (6,12?) following the subsequent release.
> 
> I think it would be helpful to the users if we could clarify our support policy when we describe the versioning change with the 1.4 release. So the key questions are:
> Q: Does everyone agree with the principle of making 1.3 a LTS release?

+1, also with my operational hat on: we'll be sticking with 1.3 for a while

> Q: Does anyone have a specific opinion on what support we should provide for our standard releases?

Given the time it has taken us in the past to release .x versions, it seems reasonable to support them for up to one year.

> 2. Should we plan a 1.5 release?
> 
> The timeline for 2.0 is currently uncertain and in particular I think the testing effort required is very hard to quantify. So with only bug fixes going into 1.4 under the new scheme, the question becomes whether we should actively plan a 1.5 release or focus solely on 2.0. 
> 
> A 1.5 release would probably consist of a number of minor developments e.g. 
> 
> https://issues.opendnssec.org/issues/?jql=project%20%3D%20OPENDNSSEC%20AND%20labels%20%3D%20%221.5%3F%22%20ORDER%20BY%20assignee%20DESC
> 
> aimed at optimising existing functionality for users of 1.4 (several being based on user requests). As long as we effectively manage the scope then this kind of release should be straightforward to plan, test and release on a defined timescale e.g. 6 months. It would mean continued minor developments on the existing enforcer code that would then needed porting to 2.0 though. For what it is worth I believe a 1.5 release would provide value to 1.4 users for a relatively small overhead and mitigate a 'big bang' feel beyond 1.4. Feedback from the board is much appreciated on this :-)

In my opinion we should focus on 2.0; past experience with co-development going on on 1.4 and 2.0 has shown that the focus tends to shift to the closer goal (1.4 in that case). That was good since 1.4 needed the attention but I feel that if we don't now pick up on 2.0 it may end up on the back burner which would be a shame given all the effort that has gone in to it so far.

Those are my two cents ;-)

Cheers,

Roland