Created
September 29, 2016 12:58
-
-
Save weiland/042a895b319a38ec2cb5a66430f401d5 to your computer and use it in GitHub Desktop.
Find or Create in Ecto
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
# two different ways to implement a find_or_create for Ecto | |
find_or_create_user = fn user -> | |
case Repo.all(from u in users, where: u.id == ^user.id and u.email == ^user.email) do | |
[] -> | |
%User{} | |
|> User.changeset(user) | |
|> Repo.insert!() | |
_ -> | |
IO.puts "Already inserted" | |
end | |
end | |
def find_or_create(user) do | |
query = from u in users, | |
where: u.uid == ^user.uid | |
if !Repo.one(query) do | |
Repo.insert(user) | |
end | |
Repo.one(query) | |
end |
Yeah, these are not safe implementations.
This is what I would do:
def find_or_create(user_params) do
{:ok, user} = %User{}
|> User.changeset(user_params)
|> Repo.insert(on_conflict: :nothing)
if is_nil(user.id) do
Repo.one from u in User, where: u.id == ^user.id
else
user
end
end
With a unique constraint on the user table (probably something like email). This uses the on_conflict
option to ensure the record exists before selecting, which should avoid race conditions: https://hexdocs.pm/ecto/Ecto.Repo.html#c:insert/2-upserts
@protestContent this is better but still not the exactly correct behavior. At least on Postgres the serial autoincrement id increases even with on_conflict: :nothing
, so with enough find_or_create
calls this will needlessly increase autoincrement IDs by quite a bit.
@wyozi good point, I didn't realize that. Still better than a race condition imo. I wonder what the postgres community recommends in cases like this.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Do these have a race condition?