On 2020-Sep-04, Laurenz Albe wrote:
> On Fri, 2020-09-04 at 10:41 -0400, Alvaro Herrera wrote:
> > > The value I see in this is:
> > > - replacing a primary key index
> > > - replacing the index behind a constraint targeted by a foreign key
> >
> > But why is this better than using REINDEX CONCURRENTLY?
>
> It is not better, but it can be used to replace a constraint index
> with an index with a different INCLUDE clause, which is something
> that cannot easily be done otherwise.
I can see that there is value in having an index that serves both a
uniqueness constraint and coverage purposes. But this seems a pretty
roundabout way to get that -- I think you should have to do "CREATE
UNIQUE INDEX ... INCLUDING ..." instead. That way, the fact that this
is a Postgres extension remains clear.
55432 14devel 24138=# create table foo (a int not null, b int not null, c int);
CREATE TABLE
Duración: 1,775 ms
55432 14devel 24138=# create unique index on foo (a, b) include (c);
CREATE INDEX
Duración: 1,481 ms
55432 14devel 24138=# create table bar (a int not null, b int not null, foreign key (a, b) references foo (a, b));
CREATE TABLE
Duración: 2,559 ms
Now you have a normal index that you can reindex in the normal way, if you need
it.
--
Álvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
