Documentation
¶
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type Scanner ¶
type Scanner[T any] struct { // contains filtered or unexported fields }
Scanner scans a result set into a struct by `db` tag while tolerating a result set whose columns do not match the struct exactly.
sqlx's StructScan is all-or-nothing in one direction: a single result column that no `db` tag claims fails every row with "missing destination name X in *T". Two things routinely produce such a column, and both are outside our control. A warehouse-managed view gains one on a vendor release, which turns an ingest into a total outage rather than a missing field. And an account configured to fold quoted identifiers to upper case returns every alias in a case the query did not ask for, so nothing matches at all.
Pinning an explicit column list in the query instead only moves the problem: the query then fails outright the day a column is withdrawn, or against a narrowed view exposing a subset.
So neither the struct nor the query decides the shape — the result set does. A column with no field is discarded and reported by UnknownColumns; a `db` tag with no column is left at its zero value and reported by MissingColumns. Callers that cannot work without a given column say so with RequireColumns rather than leaving it to the scan to fail somewhere downstream.
Only fields carrying an explicit `db` tag participate; fields the scrapper fills in itself stay untouched. Matching is case-insensitive.
func (*Scanner[T]) LogColumnDrift ¶
LogColumnDrift reports the difference between the result set and the struct, so a vendor adding or withdrawing a column surfaces in logs on the run it first happens rather than the next time someone reads the code. source names the object read, for example "SNOWFLAKE.ACCOUNT_USAGE.TASK_HISTORY"; it is left empty by callers that read an arbitrary statement rather than a named object, where the target type is what identifies the read.
func (*Scanner[T]) MissingColumns ¶
MissingColumns lists `db` tags the result set does not carry. Those fields stay at their zero value.
func (*Scanner[T]) RequireColumns ¶
RequireColumns fails when a column the caller cannot do without is absent from the result set, so an unusable read is rejected up front instead of producing rows with holes in them.
func (*Scanner[T]) UnknownColumns ¶
UnknownColumns lists result columns the struct has no field for. They are discarded rather than fatal, and are the expected shape of a vendor adding a column to a view we read.
Source Files