Container types¶

Type	Purpose
`o6.DataValue`	Value, status, and optional timestamps
`o6.ExtensionObject`	Decoded structured value or encoded payload with a type identifier
NumPy array	Typed array or matrix value

Variant¶

OPC UA Variants are implicit in Python values. There is no public o6.Variant class. The Python type selects the wire datatype; use sized o6/NumPy scalars when the width matters.

import numpy as np
import o6

value = o6.DataValue(value=o6.Int32(42))
value.value = o6.Double(3.14)
value.value = "hello"
value.value = np.array([1.0, 2.0, 3.0], dtype=np.float64)
value.value = None

A non-empty sequence infers its element type from the first element. Use a typed NumPy array for an empty array; bare [] has no inferable element type.

DataValue¶

Field	Python type	Meaning
`value`	Scalar, array, structure, or `None`	Implicit Variant payload
`status`	`o6.StatusCode`	Absent status reads as `GOOD`
`sourceTimestamp`	`o6.DateTime` or `None`	Source timestamp
`serverTimestamp`	`o6.DateTime` or `None`	Server timestamp
`sourcePicoseconds`, `serverPicoseconds`	UInt16 or `None`	Additional timestamp fractions in 10-picosecond units

import datetime
import o6

value = o6.DataValue(value=o6.Int32(42))
value.status = o6.StatusCode.BAD_TIMEOUT
value.sourceTimestamp = o6.DateTime(datetime.datetime.now(datetime.timezone.utc))
assert o6.DataValue().status == o6.StatusCode.GOOD

Read responses and data-change notifications carry DataValues. Write requests accept them; Write responses contain status codes. Method outputs are Variants unless the method declares a DataValue output.

Use client.read(..., valueOnly=False) to retain metadata. Set timestampsToReturn to select the requested timestamps.

Assigning None or deleting a DataValue field clears that field's presence.

ExtensionObject¶

Structured Python values can be assigned directly to fields that carry ExtensionObjects. An explicit wrapper exposes body and typeId; an unknown encoded type may remain an opaque payload.

import o6
from o6.ns import ns0

details = ns0.datatypes.UpdateDataDetails()
details.nodeId = o6.NodeId("ns=1;s=Temperature")
details.performInsertReplace = ns0.datatypes.PerformUpdateType.REPLACE
details.updateValues = [o6.DataValue(value=o6.Double(23.5))]

request = ns0.datatypes.HistoryUpdateRequest()
request.historyUpdateDetails = [o6.ExtensionObject(details)]

NumPy for arrays and large data¶

Decoding allocates a NumPy array and copies the native elements.
Encoding copies contiguous native-endian numeric arrays in bulk.
Strided, non-native-endian, object, and string arrays use element conversion.
Multidimensional Variant arrays preserve a valid shape; structure array fields are one-dimensional.

Returned arrays are independently owned; they are not zero-copy views of a network response. See wire-format mapping.