drop table if exists size;
create table size (
	id int primary key,
	name char(1)
);

insert into size set id=1, name='S';
insert into size set id=2, name='M';
insert into size set id=3, name='L';

drop table if exists person;
create table person (
	id int primary key,
	sid int,
	foreign key (sid) references size(id)
);

-- note: no one is wearing L
insert into person set id=1, sid=1;
insert into person set id=2, sid=1;
insert into person set id=3, sid=2;
insert into person set id=4, sid=2;
insert into person set id=5, sid=2;

-- does what i want, except leaves out sizes with totals of 0
select   s.name as size, count(s.id) as total
from     size s inner join person p on s.id=p.sid
group by s.id
order by s.id;

-- does exactly what i want, but i get the feeling there's a better way
select   s.name as size, sum(case when p.sid is null then 0 else 1 end) as total
from     size s left join (select sid from person) p on s.id=p.sid
group by s.id
order by s.id;