Re: Opptimizing projections containing unused columns

"Andrus" <kobruleht2(at)hot(dot)ee> writes:
> I have lot of autogenerated from projection queries in form
> SELECT source.c1, source.c2, t1.col1, t1.col2, ...
> FROM (SELECT c1, c2, c3, ..... , c20 FROM ... WHERE ... ) source
> LEFT JOIN t2 USING (somecolumn)
> Main SELECT uses only few columns (source.c1 and source.c2 in this sample)
> from source projection.

> 1. Does selecting of unused columns ( c3 ... c20) affect to query prefomance

Don't worry about it. All modern Postgres versions ignore columns that
aren't actually used in the query --- at least for examples as simple as
this one. In cases where you intentionally defeat optimization (eg via
OFFSET 0 in a sub-select) it's possible that the sub-select will compute
all its output columns even though the upper query doesn't use 'em all.

The width field in EXPLAIN output offers a good hint as to how many
columns the query is actually fetching.

regards, tom lane