Ignoring trailing spaces when grouping by varchar type in Redshift

I created a table in Redshift and entered data as below.

create table public.temp (
  id int,
  name varchar(20)
);

insert into public.temp (id, name) values (1, 'bob'), (2, 'bob  ');

And when counting based on the name column, it returns ('bob', 2).

select name, count(1) from public.temp group by name;

Result: ('bob', 2);

When grouping by varchar type in redshift it seems to ignore trailing spaces. And we ran another test.

select t, count(1)
from (
  select 'a' as t
  union all 
  select 'a  ' as t
) as a
group by t;

Surprisingly, it returns the result including the trailing spaces.

('a', 1), ('a  ', 1)

What is the difference between the two query execution methods and why do I get these results?