On 2/13/18 18:39, Tom Lane wrote:
> Andres Freund <andres@anarazel.de> writes:
>> Do we have a policy about catversion bumps for information schema
>> changes? A cluster from before this commit fails the regression tests
>> after the change, but still mostly works...
>
> I think historically we've not bumped catversion, on the grounds that
> there's no incompatibility with the backend as such. However, it is
> kind of annoying that not updating means the regression tests fail.
> Informally, I'm sure most developers take "catversion bump" to mean
> "you need to initdb". So I'd support saying that an information_schema
> change should include a catversion bump if it involves any changes in
> regression test results.
I will do that in the future if that is the preference.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services