]> granicus.if.org Git - postgresql/commit
Fix pg_get_indexdef()'s behavior for included index columns.
authorTom Lane <tgl@sss.pgh.pa.us>
Thu, 19 Jul 2018 17:48:05 +0000 (13:48 -0400)
committerTom Lane <tgl@sss.pgh.pa.us>
Thu, 19 Jul 2018 18:53:48 +0000 (14:53 -0400)
commit028e3da2949482fa56f3ffdbda840782f2953b0b
tree44322e3fc529acb235fa4b46cbaf80db5c6bf73a
parent1573995f55994ee04dd0d69481de17d662ad8e88
Fix pg_get_indexdef()'s behavior for included index columns.

The multi-argument form of pg_get_indexdef() failed to print anything when
asked to print a single index column that is an included column rather than
a key column.  This seems an unintentional result of someone having tried
to take a short-cut and use the attrsOnly flag for two different purposes.
To fix, split said flag into two flags, attrsOnly which suppresses
non-attribute info, and keysOnly which suppresses included columns.
Add a test case using psql's \d command, which relies on that function.

(It's mighty tempting at this point to replace pg_get_indexdef_worker's
mess of boolean flag arguments with a single bitmask-of-flags argument,
which would allow making the call sites much more self-documenting.
But I refrained for the moment.)

Discussion: https://postgr.es/m/21724.1531943735@sss.pgh.pa.us
src/backend/utils/adt/ruleutils.c
src/test/regress/expected/index_including.out
src/test/regress/sql/index_including.sql