Re: Ought to use heap_multi_insert() for pg_attribute/depend insertions?

On Sun, Nov 17, 2019 at 12:01:08AM +0100, Daniel Gustafsson wrote:
> Fixed by opting for the latter, mostly since it seems best convey what the
> function does.

- recordMultipleDependencies(depender,
- context.addrs->refs, context.addrs->numrefs,
- behavior);
+ recordMultipleDependencies(depender, context.addrs->refs,
+ context.addrs->numrefs, behavior);
Some noise diffs.

--- a/src/test/regress/expected/create_index.out
+++ b/src/test/regress/expected/create_index.out
index concur_reindex_ind4 | column c1 of table
- index concur_reindex_ind4 | column c1 of table
index concur_reindex_ind4 | column c2 of table
This removes a duplicated dependency with indexes using the same
column multiple times. Guess that's good to get rid of, this was just
unnecessary bloat in pg_depend.

The regression tests of contrib/test_decoding are still failing here:
+ERROR: could not resolve cmin/cmax of catalog tuple

Getting rid the duplication between InsertPgAttributeTuples() and
InsertPgAttributeTuple() would be nice. You would basically finish by
just using a single slot when inserting one tuple..
--
Michael