Question

Problem: I’d like to update rows that don’t match a left join (or another fast solution).
Initial goal: Update records of mytable1 that have state=0 and has either (XOR!) column “a” that matches mytable2, or column “b” that matches mytable2 (“a” and “b” should not both match!). Set the records of both tables to state=5.

I tried (and failed):

update mytable1 as t1 
    left join mytable2 as t2 on (t1.a=t2.a and t1.b=t2.b) 
set t1.state=5,t2.state=5 
where t1.state=0 and t2.state=0 and t2.a is null;

As you can see I tried to join all records that match BOTH values, so that I could update records that don’t match, also updating the rows that didn’t match in mytable2. Rows from mytable1 get updated but not those from table2. I could update the rows that DO match the left join, but that’s 99% of the rows which would be a performance hit I think (I did ask this question so I never got to compare ;).

Thank you for your time.

Answer 1

Here’s what I came up with:

update mytable1 t1 inner join mytable2 t2 on (t1.a=t2.a or t1.b=t2.b or t2.c=t1.c) set t1.state=5,t2.state=5 where((cast(t1.a=t2.a as unsigned integer) + (cast(t1.b=t2.b as unsigned integer) + (cast(t1.c=t2.c as unsigned integer)) <3) and t1.state=0;

Yes, max 2 values could be equal, only 1 should be different.

It looks like it works, and it’s fast. Any comment?