summaryrefslogtreecommitdiff
path: root/doc/src
diff options
context:
space:
mode:
authorTom Lane <tgl@sss.pgh.pa.us>2025-09-16 13:05:53 -0400
committerTom Lane <tgl@sss.pgh.pa.us>2025-09-16 13:05:53 -0400
commit8abbbbae610cd62faad4f9ec18e7cfa908593796 (patch)
tree9fc35b902d0523281aa36f91e55b8b50307b2045 /doc/src
parentc044b50d199cf592eba2ac84f48cf0fcb5fb65f8 (diff)
Revert "Avoid race condition between "GRANT role" and "DROP ROLE"".
This reverts commit 98fc31d6499163a0a781aa6f13582a07f09cd7c6. That change allowed DROP OWNED BY to drop grants of the target role to other roles, arguing that nobody would need those privileges anymore. But that's not so: if you're not superuser, you still need admin privilege on the target role so you can drop it. It's not clear whether or how the dependency-based approach to solving the original problem can be adapted to keep these grants. Since v18 release is fast approaching, the sanest thing to do seems to be to revert this patch for now. The race-condition problem is low severity and not worth taking risks for. I didn't force a catversion bump in 98fc31d64, so I won't do so here either. Reported-by: Dipesh Dhameliya <dipeshdhameliya125@gmail.com> Author: Tom Lane <tgl@sss.pgh.pa.us> Discussion: https://postgr.es/m/CABgZEgczOFicCJoqtrH9gbYMe_BV3Hq8zzCBRcMgmU6LRsihUA@mail.gmail.com Backpatch-through: 18
Diffstat (limited to 'doc/src')
-rw-r--r--doc/src/sgml/ref/drop_owned.sgml2
1 files changed, 1 insertions, 1 deletions
diff --git a/doc/src/sgml/ref/drop_owned.sgml b/doc/src/sgml/ref/drop_owned.sgml
index 46e1c229ec0..efda01a39e8 100644
--- a/doc/src/sgml/ref/drop_owned.sgml
+++ b/doc/src/sgml/ref/drop_owned.sgml
@@ -33,7 +33,7 @@ DROP OWNED BY { <replaceable class="parameter">name</replaceable> | CURRENT_ROLE
database that are owned by one of the specified roles. Any
privileges granted to the given roles on objects in the current
database or on shared objects (databases, tablespaces, configuration
- parameters, or other roles) will also be revoked.
+ parameters) will also be revoked.
</para>
</refsect1>