[postgis-devel] PostGIS 3.3.0alpha1 released

[Regina Obe]

> > > On May 27, 2022, at 12:18 PM, Sandro Santilli <strk at kbt.io> wrote:
> > >
> > > On Tue, May 24, 2022 at 12:36:46PM -0700, Paul Ramsey wrote:
> > >
> > >> why we push the patch number down into our scripts
> > >
> > > Because ALTER EXTENSION needs a <current_version>--<next_version> to
> > > do ANY upgrade at all ?
> >
> > Yes, exactly to run an upgrade SCRIPT. But if we commit to no
> > script-level changes in patch releases, then we can strip out the need
> > to ever to patch- level ALTER EXTENSION calls.
> >
> > P
> >
> >
> > >
> > > --strk;
> [Regina Obe]
> The only case we'd need them is in our master branch.
> But then as long as we keep the -next hack that satisfies that need.
[Regina Obe] 

You know thinking about this a little more,  
If people use

SELECT postgis_extensions_upgrade();

Can't we in theory force a (check the latest version, if we are at latest do
a --{latestversiongoeshere}next

--{latestversion}

Or you can't run CREATE EXTENSION twice in the function?
Since 3.3.0 isn't out yet, we could at least add a switch that says "force"
(or pass in a version number, which we wanted to do anyway) and if that is
run, it will force the upgrade with the --{latestversion}next bit