[PROJ] Backport policy to 6.0.x for database content and structure change ?

[Greg Troxel]

Even Rouault <even.rouault at spatialys.com> writes:

> Are there thoughts about if this is material appropriate for backport to 
> 6.0.x? (the backport iself is technically trivial)

seems ok

> My point of view is that generally stable branches should be for bugfixes 
> only, and not feature changes, and such changes could be considered as 
> features.

Mostly agreed, but what is really needed in a stable branch is
stability, which means a very high confidence that given any platform
where the release branch builds, when applying a change that new branch
will build and work correctly and have the same API and ABI both.  This
includes platforms on which the person applying the change cannot test.

That's of course a definition that can only be shown to be false by
example, but as a working plan it seems reasonable.

And agreed that 6.0 is itself a huge change, so more measured changes on
the way to 6.1 or whatever it is called are not so important.  This is
particularly true before packaging systems actually adopt 6 as the main
approach.   So far there are so many depending packages with issues that
I am not aware of that happening.