Enum ColumnLevelDiagnosticKind 

pub enum ColumnLevelDiagnosticKind {
    UnsupportedStatement,
    WildcardSuppressed,
    TooManyTableQualifiers,
    InsertColumnsUnresolved,
    InsertColumnsArityMismatch,
    AnonymousColumnsSuppressed,
}

Why a column-level extraction is incomplete.

Every variant is a tool-side coverage gap: sql-insight chose not to (or couldn’t) fully analyze the construct, and a more capable analyzer could do more. Per-reference resolution outcomes (ambiguous / unresolved columns) are not diagnostics — they surface on each ColumnRead::resolution instead, so the consumer reads them off the reference rather than cross-referencing a parallel diagnostic stream.

Variants

UnsupportedStatement

Statement variant the resolver / extractor does not understand well enough to extract operations from. message names the statement.

WildcardSuppressed

SELECT * / t.* left unexpanded — the extractor does not perform wildcard expansion (see crate docs), so column lineage is incomplete for projections that include a wildcard.

TooManyTableQualifiers

A table reference with more identifiers than catalog.schema.name (e.g. a SQL Server server.db.schema.table) that can’t be represented as a TableReference, so the relation — and any column read / write through it — is dropped. message names the offending identifier.

InsertColumnsUnresolved

A column-list-less INSERT / MERGE … WHEN … INSERT whose target column list couldn’t be determined without a catalog, so the source columns can’t be paired with target columns: column-level writes and lineage are dropped (the table itself still surfaces in table_writes). Supply a Catalog to resolve it. message names the target.

InsertColumnsArityMismatch

An explicit target column list whose count differs from the source query’s projected column count — an INSERT INTO t (a, b, c) <source>, or a CREATE TABLE t (a, b, c) AS / CREATE VIEW / ALTER VIEW with a column list. The positional pairing zips to the shorter side, so the surplus target columns get no lineage edge (their writes still surface) — a silent truncation this flags. message states both counts.

AnonymousColumnsSuppressed

A CREATE TABLE … AS / CREATE VIEW … (without an explicit column list) whose source projects one or more outputs with no derivable name — unaliased expressions like SELECT a + 1. The created relation does get those columns, but their names are engine-specific and not recoverable from the SQL text (e.g. PostgreSQL ?column?), so they can’t be named writes / lineage targets and are dropped. Alias the expressions to surface them. message names the relation and the count.

Trait Implementations

impl Clone for ColumnLevelDiagnosticKind

fn clone(&self) -> ColumnLevelDiagnosticKind

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ColumnLevelDiagnosticKind

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Eq for ColumnLevelDiagnosticKind

impl PartialEq for ColumnLevelDiagnosticKind

fn eq(&self, other: &ColumnLevelDiagnosticKind) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl StructuralPartialEq for ColumnLevelDiagnosticKind