@vantreeseba/drizzle-graphql 9.1.1 → 10.0.1

package/README.md

@@ -51,7 +51,7 @@ Automatically create GraphQL schema or customizable schema config fields from Dr
                 // Create a custom one
                 customUsers: {
                     // You can reuse and customize types from original schema
-                    type: new GraphQLList(new GraphQLNonNull(entities.types.UsersItem)),
+                    type: new GraphQLList(new GraphQLNonNull(entities.types.Users)),
                     args: {
                         // You can reuse inputs as well
                         where: {
@@ -79,12 +79,20 @@ Automatically create GraphQL schema or customizable schema config fields from Dr
     const yoga = createYoga({
         schema
     })
+    const server = createServer(yoga)
 
     server.listen(4000, () => {
         console.info('Server is running on http://localhost:4000/graphql')
     })
     ```
 
+    The keys on `entities` follow the naming rules below: with the defaults, table `users`
+    gives type `entities.types.Users`, inputs `entities.inputs.UsersFilters` /
+    `UsersOrderBy` / `CreateUsersInput` / `UpdateUsersInput`, queries
+    `entities.queries.users` / `usersSingle`, and mutations
+    `entities.mutations.createUsers` / `updateUsers` / `deleteUsers`. A `typeNameMapper`
+    changes these keys too, so a composed schema has to use the mapped names.
+
 ## Naming the generated types and fields
 
 By default every name comes from the table's key in the Drizzle schema: `tasks` gives type
@@ -1831,6 +1839,19 @@ image, a stored value the scalar transporting it cannot represent, is
 `DRIZZLE_UNREPRESENTABLE_VALUE`; it says the data is out of range, not that the request was
 wrong.
 
+An `update`'s `set` giving a `NOT NULL` column an explicit `null` is `DRIZZLE_NOT_NULL`, and the
+error names the column. Every field of an update input is nullable — that is what makes a partial
+`set` possible — so the nullability says nothing about the column, and an update has no default to
+fall back on: `null` there can only mean "write null". Nothing is written, including the rest of
+the `set`. To leave a column alone, omit it.
+
+An **insert** reads the same `null` the other way, and is unchanged. A `NOT NULL` column's field
+on a create input is non-null unless the database, drizzle, or a nested `create` can fill it in,
+so the only nulls that reach a resolver are ones the schema offered as "I am not supplying this" —
+and they let the default apply, which is what makes `id: $id` with a null variable insert a new
+row. A `NOT NULL` column with nothing to fill it is rejected while GraphQL coerces the request.
+Upserts take the create input, so they follow the insert rule.
+
 The `drizzle` block is only on errors raised from inside a field. An argument rejected while
 GraphQL coerces the request — a `BigInt` variable that is not an integer, say — is rejected
 before any resolver runs, so it carries its `code` but no `drizzle` and no `path`.