Virtual Fields

Make your DynamoDB datareadable and connected.

Calculate values, format raw attributes, and bring related records into your results. Inspect the records behind a value without changing what’s stored in DynamoDB.

Powered by JSONata, with presets for common calculations and live previews for your expressions.

Try the Orbit tables in Sample Mode without an AWS account.

Open all in new tab runs satelliteId = sat-corvid-2 on OrbitPasses and returns all 46 of Corvid-2’s passes, each with its station, country and outcome.

Sample ModeOrbit sample data in Dynomatic Sample Mode. This condensed demonstration follows a satellite into its related passes.

Make raw attributes easier to read. Show health as a badge, battery as a percentage, and calculated values in their own columns. Each computed column stays separate from the attributes stored in DynamoDB.

  • Computed. Not stored in DynamoDB.

One expression language, several ways to understand your data. Use JSONata to read nested attributes, calculate values, and combine data from configured links. Presets generate common expressions, while collection aggregates evaluate expressions over related records. JSONata language reference ↗

Check a calculation before you save it. Write a JSONata expression and preview its values on loaded records. Choose a display format while the preview shows the result.

  • Expression: combine the pass’s attributes.
  • Preview: inspect the result on loaded records.
  • Format: choose how the result appears.

Bring related records into the same view. Choose a target table and match its keys to each row. Add related values or summarize a collection with counts, aggregates, chips, and sparklines.

Open the records behind the result. View a complete value, preview related records, expand them beneath a row, or open their query in another tab.

  • All six ground stations behind Corvid-2’s chips
  • The first 25 related passes, 25 of 25+

Know what each total covers. Related counts and aggregates can start with the first page of records. Partial values stay marked. Compute Exact reads the remaining related records for the one cell you run it on.

  • Aurora Ridge reads 25+ passes from its first page
  • Compute Exact counts 57. Data received stays marked ≥

Details and limits

JSONata and Dynomatic
JSONata is the expression language behind Virtual Fields. Use it to extract values, calculate results, and work with related records. Presets write common expressions for you, and the preview shows what each expression returns. Dynomatic provides the items, resolves configured links, supplies additional functions, and renders the results: links decide which records an expression can read, the expression calculates, the display format draws the result, and the scope says which records it covers.
Expression contexts
An item expression reads the current item, such as batteryPct; an absent attribute gives no value. A link key expression reads the source row and must match the target key's type, and in Items from a list each element is $value. Linked data comes through $related("satellite") or $relatedAll("passes"), with the link named as a string literal; link keys cannot call either. A collection aggregate reads its related records as $items, which can be partial. Across loaded rows, expressions pick each row's value, key or group, such as bytesDown, cannot call $related or $relatedAll, and cover only loaded rows.
Bounded and complete reads
An expression can only calculate over the records its context provides. A correct formula over a partial collection still produces a partial result. Corvid-2's preview starts with 25+ passes and its full related query holds 46; Aurora Ridge's count goes from 25+ to 57 with Compute Exact. $count($distinct($relatedAll("passes").satelliteId)) reads the whole collection, which can take more reads and stops with an error past 1,000 related items or 16 MiB for a row. It never scans a table.
Calculation and format
The expression returns the value. The display format controls how the grid draws it. Battery's batteryPct returns 0.88, and Percent with fraction input and no decimals draws 88%. Downlink per minute returns about 1655.4331, and its Number format draws 1,655.4. Formats never change the stored attribute.
Preview outcomes
No value means the expression produced nothing for that item. Zero is a valid number, not no value. A compile error means the expression cannot compile, so Save is blocked. A runtime error on a row means evaluation failed there, while other rows can still give results.
Presets generate JSONata
Presets are forms that generate JSONata. Choose the attributes and operation, then inspect the generated expression and its preview. Convert to an expression when you need to customize the calculation. Converting a preset to an expression is a one-way change for that field.
Dynomatic functions
$related, $relatedAll and $gunzip are Dynomatic functions, and $items and $value are values Dynomatic supplies; none is standard JSONata. $sum, $count, $distinct and $eval are standard JSONata functions. Core temp's $eval($gunzip(telemetry)).tempC decompresses the telemetry text, interprets it, and extracts the temperature.
Stored items remain unchanged
Field definitions and computed results do not add attributes to stored DynamoDB items. Dynomatic keeps the definitions and computes the values as it shows them.
Related data requires reads
Resolving links can issue DynamoDB reads and consume capacity. Dynomatic reports Related reads separately from the read that loaded your rows.
Partial related results
When a count, a list of chips or an aggregate covers only the first related records, it keeps its marker: + after a count or a list of chips, ≥ before a value that can only grow, such as a sum of positive numbers, and ~ before any other value. Compute Exact reads the rest for the one cell you run it on.
Across loaded rows
Running totals, rankings, and shares use loaded results and can change as more rows load. Their columns say Depends on loaded rows.
Filtering computed values
Computed filters narrow loaded rows. They do not become DynamoDB query predicates, so they never change what a read requests.
Explicit links
Data Model inference and configured table links are separate mechanisms. Inference discovers possible structure from observed items. It does not create table links.
Optional AI assistance
When configured, AI can help write the expression. Review the generated expression and preview before saving. It runs with explicit consent when configured AI providers process applicable data. Describe it sends attribute names and types, and up to three preview rows only when you turn on Include sample values. Manual expressions and presets remain available.
Plans
Virtual Fields is not part of every plan. See Pricing for current availability.

Goes well with

Make sense of the datayou already have.