Struct KeyColumn 

pub struct KeyColumn<'a> { /* private fields */ }

One column of a committed table, scanned in rid order.

A rid is a row’s position in append order, and the parts of a table are in append order, so a scan of the parts in order is a scan in rid order and there is nothing to look up. That is the whole of the correspondence and it is worth stating, because a build that read the parts in any other order would produce a map that resolved every key to the wrong row without failing.

Implementations

impl<'a> KeyColumn<'a>

pub fn new(reader: &'a Reader, column: usize) -> Result<Self>

Names a column of a table as the key column of a relationship’s parent side.

Errors

If there is no such column, or if its type has no integer key form. VARCHAR is the second of those today: section 2.2 says a string key is mapped through its dictionary codes rather than its text, and the code path is not built yet. None of TPC-H’s eight relationships needs it, so it is refused by name rather than approximated.

Trait Implementations

impl<'a> Debug for KeyColumn<'a>

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

Formats the value using the given formatter.

impl Keys for KeyColumn<'_>

fn scan(&self, each: &mut dyn FnMut(Option<i128>) -> Result<()>) -> Result<()>

Calls each once per row of the column, in rid order.