using the nextval('sequence_name') in sql, the result maybe is not right

Hi, all:
PostgreSQL version : 10.3. I use "nextval" in the sql , but I think the result is not right, maybe it is a bug.
The test case as bellow:
create sequence seq1;
select nextval('seq1');
create table tx1(id1 int, id2 int);
insert into tx1 select generate_series(1,100), random()*102;
explain verbose select * from tx1 where id2 = nextval('seq1');;
select * from tx1 where id2 = nextval('seq1');
postgres=# explain verbose select * from tx1 where id2 = nextval('seq1');; QUERY PLAN ------------------------------------------------------------ Seq Scan on public.tx1 (cost=0.00..43.90 rows=11 width=8) Output: id1, id2 Filter: (tx1.id2 = nextval('seq1'::regclass)) (3 rows)

postgres=# select * from tx1 where id2 = nextval('seq1'); -- here, may be the result is not right id1 | id2 -----+----- 56 | 57 (1 row)

:: I think "nextval('seq1')" equal 2, so " select * from tx1 where id2 = nextval('seq1')" equals " select * from tx1 where id2 = 2", is it ?

Thanks,
Wanglin