Test failure with PG12
df7cb opened this issue · 2 comments
df7cb commented
pg_rational is almost compatible with PG12beta1, it just fails one test because of the new ryu float output format:
16:25:54 **** regression.diffs ****
16:25:54 diff -U3 /tmp/autopkgtest.gR9uOS/tree/expected/pg_rational_test.out /tmp/autopkgtest.gR9uOS/tree/results/pg_rational_test.out
16:25:54 --- /tmp/autopkgtest.gR9uOS/tree/expected/pg_rational_test.out 2018-05-28 05:36:50.000000000 +0000
16:25:54 +++ /tmp/autopkgtest.gR9uOS/tree/results/pg_rational_test.out 2019-05-21 14:25:53.891093629 +0000
16:25:54 @@ -76,9 +76,9 @@
16:25:54 (1 row)
16:25:54
16:25:54 select '1/3'::rational::float;
16:25:54 - float8
16:25:54 --------------------
16:25:54 - 0.333333333333333
16:25:54 + float8
16:25:54 +--------------------
16:25:54 + 0.3333333333333333
16:25:54 (1 row)
16:25:54
16:25:54 select '-1/2'::rational::float;
A possible fix would be to put SET extra_float_digits = 0;
into the test file.
A new release with the fix would be appreciated for us packagers. Thanks!
begriffs commented
pg_rational is *almost* compatible with PG12beta1, it just fails one
test because of the new ryu float output format:
Thanks for testing on the beta and letting me know about the upcoming
change.
A possible fix would be to put `SET extra_float_digits = 1;` into the
test file.
Great, I'll give that a try and if it works on both pg12 and earlier
versions, I'll create a tag and release.
df7cb commented
Fwiw, it's actually e_f_d = 0, I fixed the typo in my original posting.