product: assert audience: test-developer authority: normative

Variable System

Declaration

variables:
  supply_voltage: 12.0    # numeric
  channel: 1              # integer
  device_ready: false     # boolean
  fw_version: ""          # string

Templating

Syntax: {{variable_name}} — no dots, no expressions inside braces.

Works in: parameters values, outputs values, measurement.value, measurement.name, prompt.message, prompt.title.

DOES NOT work in: precondition, repeat.condition, enabled — use bare variable names there.

What does NOT work in templates

Output variables

Step return values are stored automatically. The outputs block renames them:

outputs:
  rail_3v3: "{{measured_voltage}}"   # store return key "measured_voltage" as "rail_3v3"

Output variables are stored in a single execution-scoped variable store — there IS NO per-runner scope. A variable set by a runner: dotnet step IS immediately available to a subsequent runner: python step.

Scope: every variable belongs to one execution

The test's variables: block, step outputs, prompt inputs, sub-sequence parameters: and values set from the debugger all belong to the execution that set them. A name is looked up in this order:

1. values written during the run (step outputs, prompts, parameters, debugger edits);
2. the test's own variables: block;
3. station-wide values (none are written by the station today).

Two executions running at the same time on a parallel station (MaxParallelExecutions > 1) never see each other's variables, even when both tests declare the same names. Everything an execution holds is removed when it ends, whatever its verdict.

Before 2.6.0 the variables: block was written to the station-wide scope. Concurrent runs that declared the same name overwrote each other, and a run could read a value that an earlier run of a different test had declared. Anything left there by those versions is cleared when the station starts.

Sub-sequence invocations start fresh

Sub-sequences share the caller's scope (see Standalone / Composite Test Pattern), but each invocation of a sub-sequence file starts from that file's own defaults, not from what its previous invocation left behind. When the same file is invoked again (once per connector, say):

• a variable the file declares in variables: that the caller did not provide goes back to its declared value;
• a variable the file's steps created (an output, a runner return value such as result) is removed, so a template reading it before a step writes it again resolves to nothing - and a measurement reading it is not acquired (see Measurement Patterns) instead of republishing the previous invocation's value;
• parameters: are then applied as before, and override declared defaults.

What the caller owns is untouched: a variable the test declares, or an earlier step set before the file was first invoked, is inherited on every invocation with whatever value it has now - including a value the sub-sequence itself updated last time (the "connect once" pattern). A variable the sub-sequence created that the caller has since overwritten is the caller's and is not reset either. Between invocations the caller can still read everything the last invocation left; the reset happens when the file is next entered.

Built-in variables

Station Configuration Variables (cfg.*)

The station configuration store (held in Orchestra) holds hardware-specific settings that differ between stations. At test start the engine reads all configuration entries for the current station, merges global and station-local values, then injects each entry as a cfg.* variable.

Station-local values WIN over global values when the same key exists in both scopes.

Always-injected key

Key naming convention

UPPER_SNAKE_CASE for hardware-address keys (e.g. DMM_VISA, PSU_COM_PORT). lower_snake_case for numeric tuning values (e.g. dmm_timeout_ms).

Usage example

- name: "Initialise DMM"
  runner: dotnet
  runner_type: net10.0
  assembly: "InstrumentDrivers.dll"
  class: "InstrumentDrivers.DmmDriver"
  method: "Connect"
  parameters:
    dmm_address: "{{cfg.DMM_VISA}}"
    dmm_timeout_ms: "{{cfg.DMM_TIMEOUT_MS}}"

A full snapshot of the merged station config IS saved as JSONB on every TestExecution row — enabling post-mortem queries about which config was active when a run failed.

Using cfg.* in a precondition

cfg.* MUST NOT be used directly in precondition:. To branch on a cfg.* value, read it into a bare variable first:

variables:
  dmm_visa: ""

setup:
  - name: "Read station config"
    runner: dotnet
    runner_type: net10.0
    assembly: "InstrumentDrivers.dll"
    class: "InstrumentDrivers.Config"
    method: "GetDmmVisa"
    parameters:
      dmm_visa: "{{cfg.DMM_VISA}}"
    outputs:
      dmm_visa: "{{result}}"

steps:
  - name: "Connect to DMM"
    precondition: "dmm_visa != ''"   # safe — bare variable
    ...

Execution Metadata Variables (exec.*)

Injected at test start from the API execution request:

Well-known tag keys

Any key not in this list gets a dynamic text input in the UI.

MES Metadata Variables (mes.*)

When a test is started through MES routing:

Additional mes.* keys may be present depending on the MES driver. Check with the system integrator for the full set available in your environment.

Namespace summary