Oninit® Grafana Data Sources — Query Editor

The two plugins expose different query editors because their underlying query languages are different. Both editors share the same template-variable interpolation and time-range macro contracts, so a dashboard that mixes panels backed by both plugins behaves consistently as the operator changes variables or pans the time range.

JSON query syntax

The oninit-json-datasource editor mirrors the MongoDB® find-document shape. A query is built from up to five fields:

The $date tag inside a filter signals a timestamp value — the plugin coerces the wrapped value into the Grafana time-series time field automatically. The Grafana time-range macros ($__from, $__to) are interpolated to millisecond-since-epoch values before the JSON document is parsed, so the listener sees a normal numeric value.

SQL syntax

The oninit-wire-datasource editor exposes a single SQL text area. The query is interpolated, then issued verbatim against the wire listener — the plugin does not rewrite the query. What the operator types is what the listener executes. For a time-series panel the first column of the result must be a timestamp (or coercible to one); subsequent numeric columns become metric series. For a table panel the column order is preserved verbatim.

Relational SQL

SELECT
    DATE(ts)        AS time,
    SUM(amount)     AS total_amount,
    COUNT(*)        AS order_count
FROM orders
WHERE ts BETWEEN $__timeFrom() AND $__timeTo()
  AND region = '$region'
GROUP BY DATE(ts)
ORDER BY DATE(ts);

Informix TimeSeries SQL routines

Informix exposes its TimeSeries data through a set of SQL routines (Calendar, GetRange, Apply, Hits, Recombine, …) that operate directly on a TimeSeries column. Because these are SQL routines, they ride the same wire-datasource editor as relational SQL:

SELECT
    sensor_id,
    Apply('SUM($value)/24', 'cal_hourly', GetRange(readings,
          $__timeFrom()::DATETIME YEAR TO FRACTION(3),
          $__timeTo()::DATETIME YEAR TO FRACTION(3)))::LVARCHAR
        AS daily_avg
FROM sensors
WHERE sensor_id = '$sensor';

Virtual Table interface

The same TimeSeries data is also reachable through Informix's Virtual Table (VT) interface, which presents a TimeSeries column as a relational view of one row per timestamp. The DBA defines the VT once via EXECUTE PROCEDURE TSCreateVirtualTab(…); panels then SELECT from it like any other table:

SELECT tstamp AS time, value
FROM v_sensor_readings
WHERE sensor_id = '$sensor'
  AND tstamp BETWEEN $__timeFrom() AND $__timeTo()
ORDER BY tstamp;

Both surfaces — routine calls and the VT — are SQL, so the choice between them is per-query ergonomic preference, not a plugin choice.

Template variables

Both plugins interpolate Grafana template variables before the query is issued. The standard variable forms ($var, ${var}, [[var]]) are recognised; format selectors (${var:csv}, ${var:pipe}, ${var:singlequote}) control how multi-value variables expand. For wire SQL, prefer ${var:singlequote} when the variable lands inside a string literal; for JSON filters the value is JSON-quoted by the plugin already, so no format selector is needed in the common case.

Time-range macros

The Grafana panel time range is exposed to both plugins as named macros that interpolate to the appropriate native form:

The $__timeFrom() / $__timeTo() SQL forms produce an Informix-native DATETIME YEAR TO FRACTION(3) literal suitable for direct comparison against an Informix DATETIME column without further casting.