lance-format/lance

Add `ReadFileError` and `DecodeColumnError` wrappers to provide better context for decoding errors

Open

#5,602 opened on 2025年12月31日

GitHub で見る
 (1 comment) (0 reactions) (0 assignees)Rust (695 forks)github user discovery
good first issue

Repository metrics

Stars
 (6,582 stars)
PR merge metrics
 (平均マージ 6d 1h) (30d で 219 merged PRs)

説明

Currently, errors that happen when decoding a column are missing information about:

  1. What column that is
  2. What file we are reading

This context is available higher up in the call stack that the main error is generate. We can add additional context by adding wrapper errors.

This approach is described nicely in https://sabrinajewson.org/blog/errors.

Basically, want to do things like:

enum DecodeError {
    NumberOutOfRange,
    ...
}

struct DecodeFieldError {
    field_name: String,
    field_id: i32,
    offset: u64,
    source: DecodeError,
}

fn decode(field: Field, data: Bytes) -> Result<_, DecodeFieldError> {
    ...
}

enum ReadFileErrorKind {
    IO(ObjectStoreError),
    Decode(DecodeFieldError),
}

struct ReadFileError {
    path: Path,
    kind: ReadFileErrorKind,
}

fn read_file(path: Path) -> Result<_, ReadFileError> {
    let data = read_data(path)?;
    for field in data.schema() {
        decode(field, data).map_err(|err: DecodeFieldError| {
            ReadFileError { path, kind: ReadFileErrorKind::Decode(err) }
        });
    }
}

And then the error stack would look like:

Error: failed to read file 'data.lance'

Caused by:
    0: failed to read field 'age' (id: 10) for row offset 10
    1: number out of range

コントリビューターガイド