Fix instability in parallel regression tests.

Commit f0c7b789a added a test case in case.sql that creates and then drops
both an '=' operator and the type it's for.  Given the right timing, that
can cause a "cache lookup failed for type" failure in concurrent sessions,
which see the '=' operator as a potential match for '=' in a query, but
then the type is gone by the time they inquire into its properties.
It might be nice to make that behavior more robust someday, but as a
back-patchable solution, adjust the new test case so that the operator
is never visible to other sessions.  Like the previous commit, back-patch
to all supported branches.

Discussion: <5983.1471371667@sss.pgh.pa.us>

diff --git a/src/test/regress/expected/case.out b/src/test/regress/expected/case.out
index 5c34af24dc184479d6e7a8bab56a47303b6ece9c..aac2fc0107c91ebb430c928da0c7201ec57faa40 100644 (file)
--- a/src/test/regress/expected/case.out
+++ b/src/test/regress/expected/case.out
@@ -305,6 +305,9 @@ SELECT * FROM CASE_TBL;
 -- the isNull flag for the case test value incorrectly became true, causing
 -- the third WHEN-clause not to match.  The volatile function calls are needed
 -- to prevent constant-folding in the planner, which would hide the bug.
+-- Wrap this in a single transaction so the transient '=' operator doesn't
+-- cause problems in concurrent sessions
+BEGIN;
 CREATE FUNCTION vol(text) returns text as
   'begin return $1; end' language plpgsql volatile;
 SELECT CASE
@@ -335,13 +338,9 @@ SELECT CASE volfoo('bar') WHEN 'foo'::foodomain THEN 'is foo' ELSE 'is not foo'
  is not foo
 (1 row)
 
+ROLLBACK;
 --
 -- Clean up
 --
 DROP TABLE CASE_TBL;
 DROP TABLE CASE2_TBL;
-DROP OPERATOR = (foodomain, foodomain);
-DROP FUNCTION inline_eq(foodomain, foodomain);
-DROP FUNCTION volfoo(text);
-DROP DOMAIN foodomain;
-DROP FUNCTION vol(text);

diff --git a/src/test/regress/sql/case.sql b/src/test/regress/sql/case.sql
index 6d1365e3f2f105bd8b1bac07bdceb95b13859c2f..1f5caf696fdd10987e7a06028c90d5cd3f965e43 100644 (file)
--- a/src/test/regress/sql/case.sql
+++ b/src/test/regress/sql/case.sql
@@ -167,6 +167,10 @@ SELECT * FROM CASE_TBL;
 -- the third WHEN-clause not to match.  The volatile function calls are needed
 -- to prevent constant-folding in the planner, which would hide the bug.
 
+-- Wrap this in a single transaction so the transient '=' operator doesn't
+-- cause problems in concurrent sessions
+BEGIN;
+
 CREATE FUNCTION vol(text) returns text as
   'begin return $1; end' language plpgsql volatile;
 
@@ -194,14 +198,11 @@ CREATE OPERATOR = (procedure = inline_eq,
 
 SELECT CASE volfoo('bar') WHEN 'foo'::foodomain THEN 'is foo' ELSE 'is not foo' END;
 
+ROLLBACK;
+
 --
 -- Clean up
 --
 
 DROP TABLE CASE_TBL;
 DROP TABLE CASE2_TBL;
-DROP OPERATOR = (foodomain, foodomain);
-DROP FUNCTION inline_eq(foodomain, foodomain);
-DROP FUNCTION volfoo(text);
-DROP DOMAIN foodomain;
-DROP FUNCTION vol(text);