Do we want a release schedule?

[RachL]

Following this slack discussion, here is my feedback to start the discussion.

At the OFN we have a regular release procedure (every Monday a new release is published).

This has several advantages:

1. The team does not have endless debate on whether now is the good time or tomorrow // which issue to include or not: time becomes the constraint. Some literature are using the image of a train: if one PR misses the train, it can take the next one.

2. Instance maintainers have a clear schedule and know when to expect new things

So on my side I don't think I can give an opinion on the frequency of the releases (aside of NOT releasing weekly 😁 ) but I'm in favor of a regular schedule we can communicate / plan on. This means releases can vary in size, but that should not be a problem.

Any thoughts?

pinging the tech working group @simonLouvet @RaggedStaff @jgaehring @lecoqlibre
but maybe this is a cross topic with ontology maintainers?

[RaggedStaff]

Thanks @RachL !

I agree this is definitely cross topic with Ontology team (although we're mostly here already 😉): ping @Alcoz !

I guess I was thinking of a monthly release schedule: e.g. we release on the 2nd Tuesday of each month.

I think I'd like to have a few other constraints, for example: we won't have a major release more than once or twice a year; i.e. at least 12 months since the last (or penultimate) major version was released. This has a couple of advantages: we ensure the team doesn't commit to too much work & if we are committing to support 2 major versions of the the standard, we ensure stabilise the standard for developers, so they can work with it, knowing breaking changes won't be appearing every month.

So I guess I have a few questions for the team:

1. Does that cadence (1 release per month) seem ok for you? Too frequent? Not frequent enough?

2. Minor Releases: could every (monthly) release be a minor release, or do we want to throttle that as well (the only reason I can think of would be to manage our workload) ?

3. How would we determine what is part of a release? Is it simply whatever has been merged in the next[Repo] branches of all of our repositories? If so - how do we determine if everything is aligned manage dependencies between repos (e.g. when ontology changes that rely on additional SKOS vocabularies have those changes also been merged in nextTaxonomies) ?