Christoph Berg

Test failure with PG15: test_show test_cancel test_cancel_pool_size ... FAILED

1

Building pgbouncer 1.17.0 against PG15 yields this: ``` gcc -O0 -g -O2 -ffile-prefix-map=/home/cbe/projects/postgresql/pgbouncer/pgbouncer=. -fstack-protector-strong -Wformat -Werror=format-security -Wall -Wextra -Wno-unused-parameter -Wno-missing-field-initializers -Wmissing-prototypes -Wpointer-arith -Wendif-labels -Wdeclaration-after-statement -Wold-style-definition -Wstrict-prototypes -Wundef -Wformat=2 -Wuninitialized -Wmissing-format-attribute...

**What happened**: ``` $ make metal make --directory=docker build-image make[1]: Entering directory '/home/myon/gardenlinux/docker' WARNING: There is no gardenlinux/slim on this box. Since this is the builder script, it will pull...

Hi, I can see in Makefile that the logic for running regression tests exists, but the directory test/ is not there. Does that exist elsewhere? Could it be added here,...

Fix build failure with PG15: Remove utils/int8.h

7

PG15 removes this header, and it doesn't seem necessary on PG10 either, so remove it unconditionally. @wieck: This is the build problem I mentioned at pgconf.eu.

parser/keywords.h is now common/keywords.h The regression expected output files need a new variant because plpgsql ERRORs will now show an extra CONTEXT line. The net diff for the two added...

This is throwing warnings otherwise that break the regression tests in 9.5+.

This adds standard "make installcheck" regression tests to pg_prioritize. Tested with PostgreSQL 9.1 .. 9.6.

Fixes assertion failures on cassert-enabled PG16. ``` LOG: GUC (PGC_REAL) pg_similarity.euclidean_threshold, boot_val=0.7, C-var=0.7 STATEMENT: CREATE EXTENSION pg_similarity; TRAP: failed Assert("check_GUC_init(variable)"), File: "./build/../src/backend/utils/misc/guc.c", Line: 4764, PID: 779990 ``` (Fix developed against...

pg_similarity crashes on PG16

1

See the build log: https://pgdgbuild.dus.dg-i.net/view/Binaries-Beta/job/pg-similarity-binaries-beta/lastFailedBuild/architecture=amd64,distribution=sid/console#console-section-9

Fix PG protocol handling on big-endian

13

Replace the homegrown swap-bytes machinery by system functions that work correctly independently from the host byte order. pgagroal_read/write_* were wrong on big-endian machines before. Problem noticed on Debian s390x where...