Can R factors be written into a ducklake?

[tomsing1]

I am trying to write a data.frame with a factor column to a ducklake, but I am getting an error - perhaps the ENUM type is not (yet?) supported?

When I write a data.frame frame to a regular duckdb database, the factor column is stored as an ENUM type.

library(duckdb)
con <- dbConnect(duckdb())

# regular duckdb table
df <- data.frame(
  letters = factor(letters[1:4])
)
dbWriteTable(con, "test", df, overwrite = TRUE)
dbGetQuery(con, "DESCRIBE test;")  # letters field is type ENUM
dbDisconnect(con)

But when I try to write the same data.frame to a ducklake, the operation fails

# ducklake
con <- dbConnect(duckdb())

dbExecute(
  con,
  "INSTALL ducklake;
  ATTACH 'ducklake:metadata.ducklake' AS my_ducklake (DATA_PATH 'data_files');
  USE my_ducklake;")
dbWriteTable(con, "demo", df)

and I get the following error:

Error in `dbSendQuery()`:
 ! rapi_prepare: Failed to prepare query CREATE  TABLE demo AS SELECT #1 FROM _duckdb_write_view_psccaemnxs
Error: Invalid Input Error: Failed to convert DuckDB type to DuckLake - unsupported type ENUM('a', 'b', 'c', 'd')
Run `rlang::last_trace()` to see where the error occurred.

I suspect that ducklake might simply not support the ENUM type, yet, as per its documentation. But I think parquet files offer a dictionary encoding, and - if I remember correctly - the arrow R package coerces R factors into dictionaries. Is there a way to do that in the context of a ducklake? Or maybe that's on the roadmap?

Many thanks for developing these awesome tools for the R community!

[krlmlr]

Thanks. I see the same behavior with plain SQL:

library(duckdb)
#> Loading required package: DBI

# ducklake
con <- dbConnect(duckdb())

dbExecute(
  con,
  "INSTALL ducklake;
  ATTACH 'ducklake:metadata.ducklake' AS my_ducklake (DATA_PATH 'data_files');
  USE my_ducklake;"
)
#> [1] 0

dbExecute(con, "CREATE TABLE demo (letters ENUM('a', 'b', 'c', 'd'))")
#> Error in `duckdb_result()` at duckdb-r/R/dbSendQuery__duckdb_connection_character.R:16:3:
#> ! rapi_execute: Failed to run query
#> Error: Invalid Input Error: Failed to convert DuckDB type to DuckLake - unsupported type ENUM('a', 'b', 'c', 'd')

^{Created on 2025-06-17 with reprex v2.1.1}

I don't see how dictionary encoding is related.

Can you please raise upstream if needed?