Timeseries join

Overview

timeseries-join FULL OUTER JOINs at least two series inputs on their time column and coalesces timestamps, retaining timestamps found on only one side.

Example configuration

time_column: timestamp                       # Default: join on this column.
inputs:                                     # Required: at least two inputs.
  - pattern: series:///sensors/temperature    # Required: example input.
    scope: air                              # Example column prefix; default null.
    range: null                             # Default: no time bounds.
    transforms: null                        # Default: no input transforms.
  - pattern: series:///sensors/humidity       # Required: example second input.
    scope: weather                          # Example column prefix; default null.
    range: null                             # Default: no time bounds.
    transforms: null                        # Default: no input transforms.

When selecting a range, its complete shape is {begin: "2024-01-01T00:00:00Z", end: "2024-01-02T00:00:00Z"}. Each bound defaults to null (unbounded); those dates are examples, not defaults.

Use and behavior

Install with pond mknod timeseries-join /derived/weather --config-path join.yaml, then inspect and query /derived/weather. Input scopes prefix value columns to avoid collisions. Optional range endpoints are RFC 3339 strings.

The source schemas and timestamp types are checked when the view is queried. This is a dynamic view, not a physical append-only series.

Full configuration

Settings

FieldAccepted typeRequiredDefaultDescription
inputsarray of Timeseries Inputyesnot specifiedList of input sources with patterns and optional time ranges
time_columnstringno"timestamp"Name of the timestamp column (defaults to "timestamp")

Time Range

Time range bounds for filtering

FieldAccepted typeRequiredDefaultDescription
beginstring | nullnonot specifiedOptional start time (ISO 8601 format)
endstring | nullnonot specifiedOptional end time (ISO 8601 format)

Timeseries Input

Input source with URL pattern and optional time range

FieldAccepted typeRequiredDefaultDescription
patternWatertown Urlyesnot specifiedURL pattern for matching time series files Supported URL schemes: - series:///pattern - Builtin TinyFS FileSeries (Parquet) - series+gzip:///pattern - Compressed FileSeries - csv:///pattern - CSV files (requires format conversion) - csv+gzip:///pattern?delimiter=; - CSV with decompression and options - excelhtml:///pattern - HydroVu HTML exports Example: series:///data/sensors/*.series
rangeTime Range | nullnonot specifiedOptional time range filter for this specific input
scopestring | nullnonot specifiedOptional scope prefix to add to all column names (e.g., "temperature" -> "BDock.temperature")
transformsarray | nullnonot specifiedOptional list of table transform factory paths to apply to this input's TableProvider. Transforms are applied in order before scope prefixes and SQL execution. Each transform is a path like "/etc/hydro_rename"

Watertown Url

Accepted value: string.