Postgresql - concat_ws like function?

Postgres 9.0 added the aggregate function string_agg() for the job:

SELECT string_agg(field1, ';') FROM tbl GROUP BY id;

The second parameter is the separator (similar to other aggregate functions).
See:

• Concatenate multiple result rows of one column into one, group by another column

The string function concat_ws() (Postgres 9.1+) does the same as MySQL's concat_ws() - when not abused as aggregate function. It's particularly useful to deal with null values:

SELECT concat_ws(';', field1, field2, field3) FROM tbl

You could even combine both to aggregate multiple columns:

SELECT id, string_agg(concat_ws(',', field1, field2, field3), ';') AS fields
FROM   tbl
GROUP  BY id;

Without array_agg (before 8.4), you can use:

SELECT array_to_string(
    ARRAY(SELECT field_lambda FROM table_lambda GROUP BY id), ';'
);

According to PostgreSQL wiki, you can emulate the PostgreSQL 8.4 array_agg function to get close to what you need.

CREATE AGGREGATE array_agg(anyelement) (
    SFUNC=array_append,
    STYPE=anyarray,
    INITCOND='{}'
);

this works for; the tag data must show that asset has several, that is, for asset 6, which has 2 tags, show them 1,2 being the table:

asest_id 1 1 1 6 6

tag_id 1 3 5 1 2

SELECT asset_tag.asset_id,asset_tag.tag_id, concat_ws(',',asset_tag.tag_id,asset_tag.tag_id) AS etiqueta

FROM public.asset_tag

--JOIN PUBLIC.imagen ON asset_tag.asset_id = imagen.asset_id

--WHERE imagen.asset_id = asset_tag.asset_id

GROUP BY asset_tag.asset_id,asset_tag.tag_id ;

Further people coming here for this problem, this method would not work with multiple columns(like concat_ws would) if you want to support multiple colums use

ARRAY_TO_STRING(ARRAY[$columns_string], 'my_delimiter').