Modern programming in Pascal

PascalABC.NET 4.0

PascalABC.NET 4.0 is the new generation of the PascalABC.NET programming system.

The key changes in version 4.0 are multitarget support for .NET Framework 4.7.2 and .NET 10, the official PascalABC.NET extension for Visual Studio Code with IntelliSense, and the new ML PascalABC.NET library for data analysis and machine learning.

Version 4.0 preserves continuity with PascalABC.NET 3.x while significantly expanding the platform and tooling capabilities of the system.

Multitarget: .NET Framework 4.7.2 + .NET 10

PascalABC.NET 4.0 introduces support for two target platforms:

.NET Framework 4.7.2 — the classic PascalABC.NET platform used by the main development environment and the existing library ecosystem.

.NET 10 — the modern .NET platform with an up-to-date runtime and standard library.

Users can choose the target platform depending on the task while keeping the same language and familiar programming model.

Modern .NET Capabilities

Supporting .NET 10 required a substantial redesign of the compiler backend.

The new target includes support for managed references and ref-like types, which are required for full compatibility with modern .NET APIs:

  • ref-return values;
  • ref-like types;
  • Span<T> and ReadOnlySpan<T>;
  • working with memory ranges without creating unnecessary copies.

For example, Span<T> makes it possible to represent a portion of an array or another contiguous memory block without allocating a new array. This is important for modern high-performance .NET APIs.

As a result, PascalABC.NET programs can directly use modern libraries and features of the .NET 10 platform.

PascalABC.NET for Visual Studio Code

An official extension for Visual Studio Code has been created for PascalABC.NET 4.0.

The extension uses a dedicated Language Server integrated with the PascalABC.NET semantic analyzer and provides:

  • code suggestions;
  • member completion after a dot;
  • Hover information about types and program elements;
  • Signature Help for procedures, functions, and methods;
  • display of compilation errors;
  • program compilation and execution;
  • support for .NET Framework 4.7.2 and .NET 10.

This means PascalABC.NET can now be used as a full-featured development environment not only in its own IDE, but also in Visual Studio Code.

PascalABC.NET in Visual Studio Code: .NET 10 target selection, IntelliSense, and a running program

PascalABC.NET for Visual Studio Code →

ML PascalABC.NET

ML PascalABC.NET is a new library for data analysis and classical machine learning in PascalABC.NET.

It provides a unified set of tools for the complete data workflow:

data loading → preprocessing → model training → evaluation → visualization

The library is implemented specifically for PascalABC.NET and does not require external machine learning libraries.

Working with Data

DataFrameABC provides typed data tables and core operations:

  • CSV loading;
  • data selection and filtering;
  • grouping;
  • table joins;
  • descriptive statistics;
  • data conversion for machine learning algorithms.

Data Preprocessing

ML PascalABC.NET includes preprocessing tools for:

  • filling missing values;
  • scaling numerical features;
  • encoding categorical features;
  • One-Hot Encoding;
  • PCA;
  • building sequences of transformations using pipelines.

Machine Learning

The library includes major classical machine learning algorithms for:

  • regression;
  • classification;
  • clustering.

These include linear and logistic models, kNN, decision trees, Random Forest, Gradient Boosting, K-Means, and DBSCAN.

It also includes evaluation metrics, cross-validation, and model validation tools.

Linear Algebra

ML PascalABC.NET includes its own numerical core with vectors and matrices, linear system solvers, and major matrix decompositions.

Datasets

The library is distributed with ready-to-use datasets for educational examples and experiments:

  • iris.csv — the classic Iris dataset;
  • mnist_small.csv — a compact version of the MNIST handwritten digit dataset;
  • moscow_housing.csv — Moscow housing data;
  • russian_cities.csv — data on Russian cities;
  • titanic_ru.csv — a Russian-language version of the Titanic dataset;
  • used_cars_price.csv — used car price data.

The last four datasets particularly emphasize the educational focus of the library: examples can be built not only on traditional English-language datasets, but also on familiar Russian real-world domains.

ML PascalABC.NET Website →

Why 4.0

The version number 4.0 reflects not a single change in the language syntax, but a transition of the entire PascalABC.NET system to a new generation.

Three major areas were introduced in one release:

Multitarget
.NET Framework 4.7.2 + .NET 10.

Visual Studio Code
A new full-featured PascalABC.NET development environment with semantic IntelliSense.

Machine Learning
A native library for data analysis and classical machine learning.

At the same time, PascalABC.NET 4.0 preserves compatibility and continuity with the existing PascalABC.NET language and its educational materials.

PascalABC.NET 4.0 combines the simplicity of modern Pascal, the mature .NET Framework environment, and the capabilities of the modern .NET 10 platform.