nautechsystems/nautilus_trader · error · anyhow::Error
Unknown data class
Error message
Unknown data class: {data_cls} What it means
During feather-to-parquet stream conversion, `convert_stream_to_data` checks the stream's data name against `is_supported_stream_data_type`. If the data class (e.g. a stream type the catalog cannot map to a known parquet writer) is unsupported, conversion bails instead of producing parquet with an unrecognizable schema. Only known stream data classes (quotes, trades, bars, deltas, etc.) can be converted.
Solutions
- Check the stream's data name against the supported types enforced by `is_supported_stream_data_type` and only convert supported streams.
- Exclude or manually migrate the unsupported stream directories instead of running bulk conversion.
- If the type should be supported, extend the conversion logic (or upgrade to a library version that supports the data class).
Defensive patterns
Strategy: try-catch
Validate before calling
// Only convert streams whose data name is in the supported set
if !supported_stream_data_types().contains(data_name) {
eprintln!("skipping unsupported stream: {data_name}");
return Ok(());
} Try / catch
match catalog.convert_stream_to_data(stream_id, data_cls, &files) {
Err(e) if e.to_string().starts_with("Unknown data class") => {
log::warn!("skipping unsupported stream data: {e}");
}
Err(e) => return Err(e),
Ok(()) => {}
} Prevention
- List stream directories and check each data name against the supported set before running bulk migration.
- Keep custom data types out of directories targeted by the automatic converter.
- After a library upgrade, re-scan legacy catalogs for data types added/renamed between versions.
When it happens
Trigger: Calling `convert_stream_to_data` for a stream directory whose data name is not one of the supported stream data types — e.g. custom/unknown record types written to the feather catalog, or a typo'd data_cls identifier in the stream metadata.
Common situations: Migrating legacy feather catalogs that contain stream types added after this converter was written; custom Nautilus stream data not in the supported set; renamed data types between versions leaving old directories with stale names.
Related errors
- Failed to concatenate stream batches
- Failed to reorder stream conversion batch
- Unsupported data type
- 16 lowercase hex bytes are valid TradeId
- A different simulation module extractor is already…
AI-assisted analysis of nautechsystems/nautilus_trader@18893faf8b (2026-09-08).
Data as JSON: /api/errors/59cb0cadc83459f7.
Report an issue: GitHub.
Appendix: source
Thrown at crates/persistence/src/backend/catalog.rs:3954
// Skip unsupported stream data types without error.
if Self::is_excluded_stream_data_type(data_cls) {
return Ok(());
}
// Convert data class name to filename (e.g., "quotes" -> "quotes")
// The data_cls should already be in the correct format (snake_case)
let data_name = to_snake_case(data_cls);
// List all feather files for this data class
let feather_files =
self.list_feather_files(subdirectory, instance_id, &data_name, identifiers)?;
if feather_files.is_empty() {
return Ok(());
}
if !Self::is_supported_stream_data_type(&data_name) {
anyhow::bail!("Unknown data class: {data_cls}");
}
// Process each feather file independently so that each file's identifier
// (instrument_id or bar_type from schema metadata) is preserved when writing
// to parquet. This matches the Python _convert_feather_table_to_parquet approach.
for file_path in feather_files {
let batches = self.read_feather_file(&file_path)?;
self.convert_feather_batches_to_parquet(
&data_name,
&file_path,
batches,
use_ts_event_for_ts_init,
)?;
}
Ok(())
}
View on GitHub (pinned to 18893faf8b)