Skip to content

Regression: All formatting options in COPY commands require format. prefix, but did not in DataFusion 36.0.0 #9716

@alamb

Description

@alamb

Describe the bug

#9604 from @metesynnada rationalized the COPY syntax to match CREATE EXTERNAL TABLE syntax which is great. However, now all options require a replicated format. string:

COPY source_table
to 'test_files/scratch/copy/table_csv_with_options'
-(format csv,
-'csv.has_header' false,
-'csv.compression' uncompressed,
-'csv.datetime_format' '%FT%H:%M:%S.%9f',
-'csv.delimiter' ';',
-'csv.null_value' 'NULLVAL');
+'format.has_header' false,
+'format.compression' uncompressed,
+'format.datetime_format' '%FT%H:%M:%S.%9f',
+'format.delimiter' ';',
+'format.null_value' 'NULLVAL');

In datafusion 36.0.0 the prefix was not needed

DataFusion CLI v36.0.0
❯ COPY (values (1)) TO 'test_files/scratch/copy/table/' (format parquet, 'compression' 'zstd(10)');

It would be nice to avoid requiring the repetition of format. string

To Reproduce

❯ COPY (select * from (values (1))) to '/tmp/foo.parquet' STORED AS PARQUET OPTIONS ('format.compression' snappy);
+-------+
| count |
+-------+
| 1     |
+-------+
1 row in set. Query took 0.037 seconds.

❯ COPY (select * from (values (1))) to '/tmp/foo.parquet' STORED AS PARQUET OPTIONS ('compression' snappy);
Invalid or Unsupported Configuration: could not find config namespace for key "compression"

Expected behavior

Both queries should do the same thing

Additional context

See discussion synnada-ai#10 (comment)

Metadata

Metadata

Labels

bugSomething isn't workingregressionSomething that used to work no longer does

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions