On Wed, Mar 02, 2022 at 05:40:00PM +0300, Aleksander Alekseev wrote:
> Hi Tom.
>
> Yeah, there's plenty of precedent for that coding if you look around.
> > I've not read the whole patch, but this snippet seems fine to me
> > if there's also an #undef at the end of the function.
>
> No, there is no #undef. With #undef I don't mind it either.
I don't see strong evidence for that pattern being wildly used with some naive
grepping:
#define for such use without undef:
POSTGRES_FDW_GET_CONNECTIONS_COLS
HEAP_TUPLE_INFOMASK_COLS
CONNECTBY_NCOLS
DBLINK_NOTIFY_COLS
PG_STAT_STATEMENTS_COLS
PG_STAT_STATEMENTS_INFO_COLS
HEAPCHECK_RELATION_COLS
PG_PARTITION_TREE_COLS
PG_STAT_GET_ACTIVITY_COLS
PG_STAT_GET_WAL_COLS
PG_STAT_GET_SLRU_COLS
PG_STAT_GET_REPLICATION_SLOT_COLS
PG_STAT_GET_SUBSCRIPTION_STATS_COLS
PG_GET_BACKEND_MEMORY_CONTEXTS_COLS
PG_GET_SHMEM_SIZES_COLS
PG_GET_REPLICATION_SLOTS_COLS
READ_REPLICATION_SLOT_COLS
PG_STAT_GET_WAL_SENDERS_COLS
PG_STAT_GET_SUBSCRIPTION_COLS
With an undef:
REPLICATION_ORIGIN_PROGRESS_COLS