Thread: pgsql: Add isolation tests for snapshot behavior in ri_triggers.c
Add isolation tests for snapshot behavior in ri_triggers.c They are to check the behavior of RI_FKey_check() and ri_Check_Pk_Match(). A test case whereby RI_FKey_check() queries a partitioned PK table under REPEATABLE READ isolation produces wrong output due to a bug of the partition-descriptor logic and that is noted as such in the comment in the test. A subsequent commit will fix the bug and replace the buggy output by the correct one. Author: Amit Langote <amitlangote09@gmail.com> Discussion: https://postgr.es/m/1627848.1636676261@sss.pgh.pa.us Branch ------ master Details ------- https://git.postgresql.org/pg/commitdiff/00cb86e75d6dd13a8bc07f79c2cdbe20d96c2d92 Modified Files -------------- src/test/isolation/expected/fk-snapshot.out | 124 ++++++++++++++++++++++++++++ src/test/isolation/isolation_schedule | 1 + src/test/isolation/specs/fk-snapshot.spec | 61 ++++++++++++++ 3 files changed, 186 insertions(+)
Hi Alvaro, On Fri, Apr 8, 2022 at 3:32 AM Alvaro Herrera <alvherre@alvh.no-ip.org> wrote: > Add isolation tests for snapshot behavior in ri_triggers.c > > They are to check the behavior of RI_FKey_check() and > ri_Check_Pk_Match(). A test case whereby RI_FKey_check() queries a > partitioned PK table under REPEATABLE READ isolation produces wrong > output due to a bug of the partition-descriptor logic and that is noted > as such in the comment in the test. A subsequent commit will fix the > bug and replace the buggy output by the correct one. Should this one also have been reverted? I wonder if it might be better to add this suite with the (yet unwritten) patch to fix the bug mentioned in the above commit message, which will likely be back-ported. -- Amit Langote EDB: http://www.enterprisedb.com