Struct ColumnIdGenerator

pub struct ColumnIdGenerator {
    existing: HashMap<Vec<Segment>, (ColumnId, DataType)>,
    next_column_id: ColumnId,
    version_id: TableVersionId,
}

Column ID generator for a new table or a new version of an existing table to alter.

Fields

existing: HashMap<Vec<Segment>, (ColumnId, DataType)>

Existing fields and their IDs and data types.

This is used for aligning field IDs between versions (ALTERs). If a field already exists and the data type matches, its ID is reused. Otherwise, a new ID is generated.

For a new table, this is empty.

next_column_id: ColumnId

The next column ID to generate, used for new columns that do not exist in existing.

version_id: TableVersionId

The version ID of the table to be created or altered.

For a new table, this is 0. For altering an existing table, this is the next version ID of the version_id field in the original table catalog.

Implementations

impl ColumnIdGenerator

pub fn new_alter(original: &TableCatalog) -> Self

Creates a new ColumnIdGenerator for altering an existing table.

pub fn new_initial() -> Self

Creates a new ColumnIdGenerator for a new table.

pub fn generate(&mut self, col: &mut ColumnCatalog) -> Result<(), RwError>

Generate ColumnIds for the given column and its nested fields (if any) recursively. The IDs of nested fields will be reflected in the updated DataType of the column.

Returns an error if there’s incompatible data type change.

pub fn into_version(self) -> TableVersion

Consume this generator and return a TableVersion for the table to be created or altered.

Trait Implementations

impl Debug for ColumnIdGenerator

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

Formats the value using the given formatter.