Output

Save data from your workflow to files.

Sockets

Socket	Direction	Description
input	Input	Data to save

The Output tool has no output sockets - it's a terminal node in the workflow.

Supported Formats

CSV

Comma-separated values and other delimited text files.

Writing Modes:

Mode	Description	Best For
Sink	Streaming write, memory efficient	Large datasets
Write	Collects all data first, then writes	When you need specific write options

CSV Options:

Option	Default	Description
Destination	(required)	Path for the output file
Separator	,	Field delimiter
Include Header	true	Write column names as first row
Include BOM	false	Write UTF-8 byte-order mark at start of file
Quote Character	"	Character used to quote fields
Quote Style	necessary	When to quote fields
Line Terminator	\n	End-of-line character(s)
Null Value	(empty)	String to represent null values
Date Format	(ISO)	Format for date columns
Datetime Format	(ISO)	Format for datetime columns
Time Format	(ISO)	Format for time columns
Float Precision	(full)	Decimal places for floats
Float Scientific	(auto)	Use scientific notation for floats

Parquet

Columnar binary format, efficient for analytics.

Writing Modes:

Mode	Description	Best For
Sink	Streaming write	Large datasets
Write	Collects all data first	When you need specific write options

Parquet Options:

Option	Default	Description
Destination	(required)	Path for the output file
Compression	zstd	Compression algorithm (zstd, snappy, lz4, gzip, none)
Compression Level	(default)	Compression level (algorithm-specific)
Row Group Size	(default)	Rows per row group
Data Page Size	(default)	Size of data pages within row groups
Statistics	true	Write column statistics

IPC / Arrow

Apache Arrow's native binary format (.arrow, .ipc, .feather). Fastest format for read/write operations.

Writing Modes:

Mode	Description	Best For
Sink	Streaming write	Large datasets, memory efficiency
Write	Collects all data first	Small datasets

IPC Options:

Option	Default	Description
Destination	(required)	Path for the output file (.arrow, .ipc, or .feather)
Compression	uncompressed	Compression (uncompressed, lz4, or zstd)
Compat Level	(newest)	Arrow IPC compatibility level for older consumers

When to Use IPC/Arrow

• Intermediate files: Use IPC for temporary data between pipeline stages
• Cross-application sharing: IPC works with DuckDB, Spark, R, Julia, and other Arrow-compatible tools
• Maximum speed: IPC has the fastest read/write performance of any format
• Memory efficiency: Files can be memory-mapped for efficient access

Excel

Microsoft Excel spreadsheet files (.xlsx).

Writing Mode:

Excel only supports eager writing (data is collected before writing). There is no streaming sink support.

Excel Options:

Option	Default	Description
Destination	(required)	Path for the output file (.xlsx)
Worksheet	Sheet1	Name of the worksheet to create
Table Name	(empty)	Optional name for the Excel table object
Table Style	Table Style Medium 4	Excel table style (e.g., "Table Style Medium 9")
Include Header	true	Write column names as header row
Autofilter	true	Enable autofilter dropdown in header
Autofit	false	Auto-calculate column widths to fit content
Hide Gridlines	false	Hide worksheet gridlines
Float Precision	3	Decimal places for float numbers
Sheet Zoom	100%	Zoom level (10-400%)
Freeze Panes	(none)	Cell position to freeze (e.g., "A2" freezes top row)

Excel Table Styles

Excel table styles provide professional formatting. Popular options include:

• Light styles (1-21): Subtle, minimalist
• Medium styles (1-28): Balanced, good for most uses
• Dark styles (1-11): High contrast, bold headers
• None: Plain data without table formatting

You can also enter any custom Excel table style name. See the full visual reference below.

Click to view all Excel table styles

Avro

Apache Avro binary format (.avro). Row-based format for big data ecosystems.

Writing Mode:

Avro only supports eager writing (data is collected before writing). There is no streaming sink support.

Avro Options:

Option	Default	Description
Destination	(required)	Path for the output file (.avro)
Compression	uncompressed	Compression (uncompressed, snappy, or deflate)
Schema Name	(empty)	Optional name for the Avro schema

Avro Use Cases

Use Avro when:

• Integrating with Hadoop, Kafka, or Spark pipelines
• Schema evolution is important for your data workflow
• Interoperability with Java/JVM ecosystems is required

For maximum performance within Sigilweaver Loom, prefer IPC/Arrow or Parquet.

JSON / NDJSON

JSON data files. The format is auto-detected by file extension.

Format	Extensions	Writing Mode	Description
Standard JSON	.json	Write (eager)	Array of objects or column-oriented
NDJSON	.jsonl, .ndjson	Sink (streaming) or Write	Newline-delimited JSON

JSON Options (standard .json):

Option	Default	Description
Destination	(required)	Path for the output file
Row Oriented	true	Write as array of row objects (vs column-oriented)

NDJSON Options (.jsonl, .ndjson):

Option	Default	Description
Destination	(required)	Path for the output file

NDJSON for Large Files

Use NDJSON format (.jsonl or .ndjson) for streaming sink support. Standard .json files must be collected into memory before writing.

YXDB

YXDB database files (.yxdb). Written using OpenYXDB.

Writing Mode:

YXDB only supports eager writing (data is collected before writing).

YXDB Options:

Option	Default	Description
Destination	(required)	Path for the output file (.yxdb)

Configuration

1. Add an Output tool to the canvas
2. Connect an input wire from an upstream tool
3. Select the output format (CSV, Excel, Parquet, IPC/Arrow, JSON/NDJSON, Avro, or YXDB)
4. Choose the destination path
5. Configure format-specific options as needed

Examples

Saving to CSV

1. Wire data into the Output tool
2. Select "CSV" as the output format
3. Enter the destination file path (e.g., /path/to/output.csv)
4. Adjust options like separator or header as needed

Saving to Parquet

1. Wire data into the Output tool
2. Select "Parquet" as the output format
3. Enter the destination file path (e.g., /path/to/output.parquet)
4. Choose compression (zstd recommended for balance of speed and size)

Saving to IPC/Arrow

1. Wire data into the Output tool
2. Select "IPC / Arrow" as the output format
3. Enter the destination file path (e.g., /path/to/output.arrow)
4. Choose compression if needed (uncompressed is fastest, zstd for smaller files)

Saving to Excel

1. Wire data into the Output tool
2. Select "Excel" as the output format
3. Enter the destination file path (e.g., /path/to/output.xlsx)
4. Optionally configure worksheet name, table styling, and formatting options
5. Use freeze panes to keep headers visible when scrolling

Saving to Avro

1. Wire data into the Output tool
2. Select "Avro" as the output format
3. Enter the destination file path (e.g., /path/to/output.avro)
4. Choose compression (uncompressed for speed, snappy or deflate for smaller files)

Saving to JSON

1. Wire data into the Output tool
2. Select "JSON" as the output format
3. Enter the destination file path:
   • Use .json for standard JSON (array of objects)
   • Use .jsonl or .ndjson for newline-delimited JSON (streaming support)
4. For standard JSON, choose row-oriented (array of objects) or column-oriented output

Saving to YXDB

1. Wire data into the Output tool
2. Select "YXDB" as the output format
3. Enter the destination file path (e.g., /path/to/output.yxdb)

Notes

• Destination directories are created automatically if they don't exist
• Sink mode is recommended for large datasets to minimize memory usage (CSV, Parquet, IPC, NDJSON)
• Excel, Avro, standard JSON, and YXDB do not support streaming - data is collected before writing
• IPC/Arrow is the fastest format - use it when speed is critical
• Excel files are best for business reporting and sharing with non-technical users
• YXDB files are supported via OpenYXDB
• NDJSON (.jsonl, .ndjson) supports streaming sink unlike standard .json files
• Overwriting: Existing files at the destination path will be overwritten
• Execution: Output tools only write files when the workflow is executed (not during preview)