Simplify reindexdb coding get_parallel_object_list() was trying to serve two masters, and it was doing a bad job at both. In particular, it treated the given user_list as an output argument, but only sometimes. This was confusing, and the two paths through it didn't really have all that much in common, so the complexity wasn't buying us much. Split it in two: get_parallel_tables_list() handles the straightforward cases for schemas, databases and tables, takes one list as argument and returns another list.
A new function get_parallel_tabidx_list() handles the case for indexes. This takes a list as argument and outputs two lists, just like get_parallel_object_list used to do, but now the API is clearer (IMO anyway). Another difference is that accompanying the list of indexes now we have a list of tables as an OID list rather than a fully-qualified table name list. This makes some comparisons easier, and we don't really need the names of the tables, just their OIDs. (This requires atooid, which requires <stdlib.h>). Author: Ranier Vilela <ranier...@gmail.com> Author: Álvaro Herrera <alvhe...@alvh.no-ip.org> Discussion: https://postgr.es/m/CAEudQArfqr0-s0VVPSEh=0kgOgBJvFNdGW=xsl5rbcr0wdm...@mail.gmail.com Branch ------ master Details ------- https://git.postgresql.org/pg/commitdiff/f76892c9ff7e5f2dcb4073310d1a5273f47d1d9d Modified Files -------------- src/bin/scripts/reindexdb.c | 275 +++++++++++++++++++------------------ src/bin/scripts/t/090_reindexdb.pl | 2 + 2 files changed, 141 insertions(+), 136 deletions(-)
