compiler: Support calling functions with defaults #635
Conversation
Fix a number of bugs when resolving function calls. Switch to using functions generated from a default PostgreSQL instance. Add a new test case ripped from the PostgreSQL docs.
| @@ -11,9 +11,9 @@ const makeIntervalDays = `-- name: MakeIntervalDays :one | |||
| SELECT make_interval(days => $1::int) | |||
| ` | |||
|
|
|||
| func (q *Queries) MakeIntervalDays(ctx context.Context, dollar_1 int32) (interface{}, error) { | |||
| func (q *Queries) MakeIntervalDays(ctx context.Context, dollar_1 int32) (int64, error) { | |||
There was a problem hiding this comment.
make_interval now returns an actual type instead of interface{}. This happened because we can now resolve function calls with defaults and named parameters.
| @@ -33,9 +33,9 @@ const makeIntervalSecs = `-- name: MakeIntervalSecs :one | |||
| SELECT make_interval(secs => $1) | |||
| ` | |||
|
|
|||
| func (q *Queries) MakeIntervalSecs(ctx context.Context, secs interface{}) (interface{}, error) { | |||
| func (q *Queries) MakeIntervalSecs(ctx context.Context, secs float64) (int64, error) { | |||
There was a problem hiding this comment.
The compiler is now smart enough to pull out type information for named parameters in function calls.
| } | ||
|
|
||
| func (q *Queries) Plus(ctx context.Context, arg PlusParams) (int32, error) { | ||
| row := q.db.QueryRowContext(ctx, plus, arg.B, arg.A) | ||
| row := q.db.QueryRowContext(ctx, plus, arg.A, arg.B) |
There was a problem hiding this comment.
!!!
The compiler generated incorrect code here. The order of the arguments wasn't correct.
| @@ -17,15 +17,15 @@ type GenerateSeriesParams struct { | |||
| Column2 time.Time `json:"column_2"` | |||
| } | |||
|
|
|||
| func (q *Queries) GenerateSeries(ctx context.Context, arg GenerateSeriesParams) ([]interface{}, error) { | |||
| func (q *Queries) GenerateSeries(ctx context.Context, arg GenerateSeriesParams) ([]string, error) { | |||
There was a problem hiding this comment.
Goodbye interface{}! Real return values for function calls
| @@ -11,8 +11,8 @@ const advisoryLock = `-- name: AdvisoryLock :many | |||
| SELECT pg_advisory_unlock($1) | |||
| ` | |||
|
|
|||
| func (q *Queries) AdvisoryLock(ctx context.Context, key int64) ([]bool, error) { | |||
| rows, err := q.db.QueryContext(ctx, advisoryLock, key) | |||
| func (q *Queries) AdvisoryLock(ctx context.Context, pgAdvisoryUnlock int64) ([]bool, error) { | |||
There was a problem hiding this comment.
While the PostgreSQL documentation claims the first parameter to pg_advisory_unlock is named key, that name isn't actually included in the definition of the function stored in the database.
| @@ -5,7 +5,7 @@ import "github.com/kyleconroy/sqlc/internal/sql/catalog" | |||
| func NewCatalog() *catalog.Catalog { | |||
| c := catalog.New("public") | |||
| c.Schemas = append(c.Schemas, pgTemp()) | |||
| c.Schemas = append(c.Schemas, pgCatalog()) | |||
| c.Schemas = append(c.Schemas, genPGCatalog()) | |||
There was a problem hiding this comment.
This is a bit scary! I'm curious if we'll have to walk this specific change back.
Fix a number of bugs when resolving function calls.
Switch to using functions generated from a default PostgreSQL instance.
Add a new test case ripped from the PostgreSQL docs.
I've noted the improvements in individual comments