Question

I’m creating a db schema that involves users that can be friends, and I was wondering what is the best way to model the ability for these users to have friendships. Should it be its own table that simply has two columns that each represents a user? Thanks!

Answer 1

create table 
friendship(
user bigint, 
friend bigint,
primary key(user, friend),
key(friend, user),
constraint `fk_user` foreign key (user) references user(id),
constraint `fk_friend` foreign key (friend) references user(id)
);

When user 1 sends a friendship request to user 2, do

insert into friendship (user, friend) values (1,2);

If user 2 denies the request,

delete from friendship where user = 1 and friend = 2;

if user 2 accepts it:

insert into friendship (user, friend) values (2,1);

Then, a friendship can be found this way:

select f1.* 
from friendship f1
inner join friendship f2 on f1.user = f2.friend and f1.friend = f2.user;

You can make a view with this last query, it will help you query-ing for users’ friends, or even friends of friends.