It took a long time coming, but MySQL version 8.0.14 finally added support for lateral joins - the official terminology is lateral derived tables.
This is a very powerful feature, that comes handy in multiple situations, including unpivoting table columns to rows.
You can phrase the query as follows:
select t.id, x.*
from mytable t
cross join lateral (
select a, 'a'
union all select b, 'b'
union all select c, 'c'
) as x(col1, col2)
It may look like this is not a big difference compared to the typical cannonical solution - after all, we are still using union all
within the lateral derived table... But don't get it wrong: this query scans the table only once, as opposed to the other approach, which requires one scan for each column to unpivot. So this is more efficient - and the performance gain increases dramatically as the table goes bigger and/or more columns need to be unpivoted.
Bottom line: if you are running MySQL 8.0.14 or higher, just use this technique. From that version onwards, this is the canonical way to unpivot in MYSQL.
Demo on DB Fiddle:
Sample data:
ID | a | b | c
-: | :- | :- | :-
1 | a1 | b1 | c1
2 | a2 | b2 | c2
Query results:
id | col1 | col2
-: | :--- | :---
1 | a1 | a
1 | b1 | b
1 | c1 | c
2 | a2 | a
2 | b2 | b
2 | c2 | c
Side note
MySQL 8.0.19 added support for the VALUES
statement, which could help further shortening the query by removing the need to use union all
in a subquery (although I don't see any performance gain here, this makes the query neater).
nfortunately, As of version 8.0.21, this does not work yet - which might be considered a bug - but maybe will in a future version...:
select t.id, x.*
from mytable t
cross join lateral (values
row(a, 'a'),
row(b, 'b'),
row(c, 'c')
) as x(col1, col2);