IronBarcode Architecture Overview — As of Version 2025.2+

Overview

IronBarcode package architecture has been restructured for better modularity, cross-platform support, and long-term maintainability. The new design aligns with the naming conventions of IronOCR and IronPDF, and introduces clearer separation of C#, native, and machine learning components.

Package Structure

Main Distribution

• BarCode
  Default package for Windows platforms. This is the primary NuGet for general-purpose use on Windows.

• BarCode.Linux, BarCode.MacOS, etc.
  OS-specific distributions. These will follow a naming convention similar to the approach already used in IronOCR and IronPDF.
  

Internal Package Layers

Each top-level Barcode package is composed of multiple internal subpackages:

1. BarCode.Slim

• Contents:

  • Core C# code

  • The interop layer

  • PDF-related code

  • All C# third-party dependencies

• Purpose: Acts as the central logic and binding layer that drives IronBarcode’s managed code features.

2. IronSoftware.ReaderInternals.XXX

• Contents:

  • All native C++ code for:

    • Reading

    • Writing

    • Detecting barcodes

• Platform-specific: Variants exist for Windows, Linux, Mac, etc.

3. BarCode.Detection

• Contents:

  • Machine learning detection logic

  • Currently implemented in C# using ONNX Runtime

  • Will not be dependant on OS at the moment

• Note: Will eventually be replaced by ncnn (a native C++ ML inference library), which will introduce OS-specific variants.

Exclusions for Mobile Platforms

Due to performance concerns and compatibility limitations on mobile platforms:

• Barcode.Detection will NOT be included in:

  • Barcode.iOS

  • Barcode.Android

Benefits of the New Architecture

• Modular separation between managed and native code

• Clean OS-targeted packaging

• Future-proofing for high-performance ML detection

• Better alignment with other IronSoftware libraries

• Improved maintainability and troubleshooting