Horton

Thread

Commits

  1. Stabilize timetz test across DST transitions.

  1. BUG #16689: timetz test fails 2020-11-01

    The Post Office <noreply@postgresql.org> — 2020-10-29T17:54:03Z

    The following bug has been logged on the website:

Bug reference:      16689
Logged by:          Bernhard M. Wiedemann
Email address:      githubbmw2020@lsmod.de
PostgreSQL version: 12.4
Operating system:   openSUSE-Tumbleweed
Description:        

timetz test fails after 2020-11-01
likely from upcoming daylight saving time changes in the US

 diff -U3
/home/abuild/rpmbuild/BUILD/postgresql-12.4/src/test/regress/expected/timetz.out
/home/abuild/rpmbuild/BUILD/postgresql-12.4/src/test/regress/results/timetz.out
 ---
/home/abuild/rpmbuild/BUILD/postgresql-12.4/src/test/regress/expected/timetz.out
  2020-08-10 21:15:53.000000000 +0000
 +++
/home/abuild/rpmbuild/BUILD/postgresql-12.4/src/test/regress/results/timetz.out
   2035-11-30 21:20:50.205175526 +0000
 @@ -94,25 +94,25 @@
  SELECT '23:59:59.999999'::timetz;
         timetz
  --------------------
 - 23:59:59.999999-07
 + 23:59:59.999999-08
  (1 row)

  SELECT '23:59:59.9999999'::timetz;  -- rounds up
     timetz
  -------------
 - 24:00:00-07
 + 24:00:00-08
  (1 row)

  SELECT '23:59:60'::timetz;  -- rounds up
     timetz
  -------------
 - 24:00:00-07
 + 24:00:00-08
  (1 row)
         
  SELECT '24:00:00'::timetz;  -- allowed
     timetz
  -------------
 - 24:00:00-07
 + 24:00:00-08
  (1 row)

  SELECT '24:00:00.01'::timetz;  -- not allowed
 + exit 1


To reproduce on Debian or openSUSE:
osc checkout openSUSE:Factory/postgresql12 && cd $_
osc build --vm-type=kvm --noservice --build-opt=--vm-custom-opt="-rtc
base=2020-11-01T08:00:00" standard

Background:
As part of my work on reproducible builds for openSUSE, I check that
software still gives identical build results in the future.
This showed up failing tests in our package build. Then I bisected the
failure date.

  2. Re: BUG #16689: timetz test fails 2020-11-01

    Tom Lane <tgl@sss.pgh.pa.us> — 2020-10-29T19:10:23Z

    PG Bug reporting form <noreply@postgresql.org> writes:
> timetz test fails after 2020-11-01
> likely from upcoming daylight saving time changes in the US

Hmm, those test cases were just added a couple months ago, so they
have not survived a DST transition yet.  Thanks for the heads-up!

			regards, tom lane