Re: pgsql: RLS refactoring

Stephen Frost <sfrost(at)snowman(dot)net> writes:
> * Tom Lane (tgl(at)sss(dot)pgh(dot)pa(dot)us) wrote:
>> It looks to me like this changed the representation of stored rules, so it
>> should have included a catversion bump. This is particularly relevant to
>> the 9.5 branch where people already have alpha installations.

> I had considererd if a bump was needed and figured it wasn't.

> I don't mind doing a bump if we feel it's necessary and maybe I'm
> missing that there's a way to cause that node type to end up in the
> catalog, but I don't think so, as we only ever build WithCheckOption
> nodes in the rewriter.

Oh, I see. In that case you should remove WithCheckOption from the set of
node types supported by readfuncs.c, both because it's dead code and to
clarify that the node is not meant to ever end up on disk.

(outfuncs.c support is useful for debugging though, so keep that.)

regards, tom lane