On 12/26/19 6:46 PM, Tom Lane wrote:
Awhile back I wrote:
Actually ... maybe we don't need to change the view definition at all,
but instead just make has_column_privilege() do something different
for indexes than it does for other relation types. It's dubious that
applying that function to an index yields anything meaningful today,
so we could redefine what it returns without (probably) breaking
anything. That would at least give us an option to back-patch, too,
though the end result might be complex enough that we don't care to
In hopes of resurrecting this thread, here's a draft patch that does
it like that (and also fixes row_security_active(), as otherwise this
probably creates a security hole in pg_stats).
Do you know when you will have a chance to look at this patch?
Tom made a suggestion up-thread about where the regression tests could go.