Python and .NET Integration

Bidirectional integration enabling Python's data science ecosystem within .NET's FrameworX

Platform→ Python and .NET Integration | Tutorial | Reference

Integration Philosophy: FrameworX provides true bidirectional integration between Python and .NET, not just Python scripting capability. This design leverages Python's extensive data science libraries while maintaining .NET's performance and reliability for industrial control.

The platform requires Python 3.7 or later, delivering significant performance improvements over Python 2. According to Python.org benchmarks, Python 3 runs approximately 10-30% faster than Python 2 in most scenarios, with some operations seeing 5x improvements.

Language Selection Guide

Language	Best For	Performance	Key Features
C#	Complex logic, performance-critical operations	Excellent	Full .NET access, strongly typed, IntelliSense
Python	Data analysis, ML integration	Good	Extensive libraries, NumPy/Pandas support

Quick Selection Criteria

C# - Maximum performance, real-time control, type safety
Python - Data science, machine learning, rapid prototyping
Both - C# for control logic, Python for analytics

Integration Patterns

1. Embedded Python Scripts

Python code runs within FrameworX's Script Tasks and Classes, with full access to tags and namespaces:

# Python accessing FrameworX tags
@Tag.Temperature = @Tag.SetPoint + 5
result = ProcessData(@Tag.ProductionRate)

2. Shell Integration

Execute external Python files while maintaining tag access:

result = TK.ExecutePythonShell("analytics.py", [@Tag.Input1, @Tag.Input2])
@Tag.Result = result

Test Playground

When the editor is used to edit Script Tasks, you can test the scripts directly, using the Play button at the Task name line. The results of your code will show in the standard Designer output.

3. Cross-Language Calls

C# calling Python:

// C# code calling Python class
var result = @Script.Class.PythonAnalytics.ProcessML(data);

Python calling C#:

# Python code calling C# method
result = @Script.Class.CSharpLogic.Calculate(param1, param2)

Direct Access Architecture

Python code accesses FrameworX objects directly through the same '@' notation used in C#:

@Tag - Real-time tag values
@Dataset - Database queries
@Alarm - Alarm states
@Historian - Time-series data

No marshaling, conversion, or intermediate layers, Python operates on the same objects as .NET code.

Development Environment

Solution-Specific Python

Each solution specifies its Python interpreter, enabling:

Different Python versions per project
Isolated package environments
Consistent deployment across development and production

In-Designer Features

Code Editor - Python syntax highlighting and IntelliSense
Testing - Run Python scripts directly in Designer
Debugging - Output window for print statements and errors
Tag Access - Auto-completion for all namespace objects

Industrial Use Cases

Machine Learning Integration

# Use scikit-learn for anomaly detection
from sklearn.ensemble import IsolationForest
model = IsolationForest()
data = GetHistorianData(@Tag.Equipment)
anomalies = model.fit_predict(data)

Data Analysis

# Pandas for production analytics
import pandas as pd
df = pd.DataFrame(@Dataset.Query("SELECT * FROM Production"))
daily_avg = df.groupby('Day')['Output'].mean()
@Tag.DailyAverage = daily_avg.iloc[-1]

Custom Protocols

# Implement proprietary protocol
import struct
packet = struct.pack('>HH', @Tag.DeviceID, @Tag.Command)
response = SendCustomProtocol(packet)
@Tag.Response = struct.unpack('>H', response)[0]

System Requirements

Python Version: 3.7 or later (3.11 recommended)
Python.NET: Required for integration (pip install pythonnet)
Installation: System-wide (all users) for service deployment
Path: Python.exe must be in system PATH

Performance Considerations

Operation Type	Recommended	Reason
Real-time control loops	C#	Deterministic execution, minimal latency
Statistical analysis	Python	NumPy/Pandas optimized C libraries
Tag calculations	C#	Direct memory access, no interpreter
Machine learning inference	Python	Scikit-learn, TensorFlow ecosystems
Database operations	Either	Both have excellent SQL support

Key Advantages

No Bridge Required - Direct integration via Python.NET
Shared Memory Space - Zero-copy data access between languages
Production Ready - Runs as Windows service, handles 24/7 operation
Version Flexibility - Multiple Python versions on same server
Library Access - Full Python Package Index (PyPI) availability