adds update function

README.md

@@ -445,3 +445,106 @@ end
 ```
 
 `mix test` again, and we should be all green.
+
+#### 4.3 Update
+
+Now we come to the update function. "But I thought we were only appending to the database?" I hear you ask. This is true, but we still need to relate our existing data to the new, updated data we add.
+
+To do this, we need to be able to reference the previous entry somehow. The simplest (conceptually) way of doing this is to provide each entry a unique id. Note that the id will be used to represent unique entries, but it will not be unique in a table, as revisions of entries will have the same id. This is the simplest way we can link our entries, but there may be some disadvantages, which we'll look into later.
+
+So, first we'll need to edit our schema to add a shared id to our address entries:
+
+``` elixir
+defmodule Append.Address do
+  ...
+  schema "addresses" do
+    ...
+    field(:entry_id, :integer)
+
+    ...
+  end
+  ...
+end
+```
+
+And create a migration file:
+
+```
+mix ecto.gen.migration add_entry_id
+```
+
+``` elixir
+defmodule Append.Repo.Migrations.AddEntryId do
+  use Ecto.Migration
+
+  def change do
+    alter table("addresses") do
+      add :entry_id, :integer
+    end
+
+  end
+end
+```
+
+```
+mix ecto.migrate
+```
+
+Now, we'll write a test for the update function.
+
+``` elixir
+defmodule Append.AddressTest do
+  ...
+  test "update item in database" do
+    {:ok, item} = insert_address()
+
+    {:ok, updated_item} = Address.update(item, %{tel: "0123444444"})
+
+    assert updated_item.name == item.name
+    assert updated_item.tel != item.tel
+  end
+  ...
+end
+```
+
+Then we write the update function itself. We'll take the existing item, and a map of updated attributes as arguments.
+
+Then we'll use the `insert` function to create a new entry in the database, rather than `update`, which would overwrite the old one.
+
+``` elixir
+defmodule Append.AppendOnlyLog do
+  ...
+  defmacro __before_compile__(_env) do
+    quote do
+      ...
+      def update(%__MODULE__{} = item, attrs) do
+        item
+        |> __MODULE__.changeset(attrs)
+        |> Repo.insert()
+      end
+    end
+  end
+end
+```
+
+But now, if we try to run our tests:
+
+```
+1) test update item in database (Append.AddressTest)
+     test/append/address_test.exs:25
+     ** (Ecto.ConstraintError) constraint error when attempting to insert struct:
+
+         * unique: addresses_pkey
+```
+We can't add the item again, because the unique id already exists in the database.
+
+``` elixir
+def update(%__MODULE__{} = item, attrs) do
+  item
+  |> Map.put(:id, nil)
+  |> __MODULE__.changeset(attrs)
+  |> Repo.insert()
+end
+```
+
+So here we remove the original autogenerated id from the existing item, preventing us from duplicating it in the database.

lib/append/address.ex

@@ -10,6 +10,7 @@ defmodule Append.Address do
     field(:name, :string)
     field(:postcode, :string)
     field(:tel, :string)
+    field(:entry_id, :integer)
 
     timestamps()
   end

lib/append/append_only_log.ex

@@ -38,6 +38,10 @@ defmodule Append.AppendOnlyLog do
       end
 
       def update(%__MODULE__{} = item, attrs) do
+        item
+        |> Map.put(:id, nil)
+        |> __MODULE__.changeset(attrs)
+        |> Repo.insert()
       end
     end
   end

priv/repo/migrations/20180914130516_add_entry_id.exs

@@ -0,0 +1,10 @@
+defmodule Append.Repo.Migrations.AddEntryId do
+  use Ecto.Migration
+
+  def change do
+    alter table("addresses") do
+      add :entry_id, :integer
+    end
+
+  end
+end

test/append/address_test.exs

@@ -22,6 +22,15 @@ defmodule Append.AddressTest do
     end
   end
 
+  test "update item in database" do
+    {:ok, item} = insert_address()
+
+    {:ok, updated_item} = Address.update(item, %{tel: "0123444444"})
+
+    assert updated_item.name == item.name
+    assert updated_item.tel != item.tel
+  end
+
   def insert_address do
     Address.insert(%{
       name: "Thor",