Changelog

Version history and roadmap for the T programming language.

Version 0.1.0 (Alpha) — Current Release

Status: Alpha — Syntax and semantics frozen
Release Date: February 2026

Language Core

✅ Implemented: - Lexer and parser (Menhir) - Tree-walking interpreter - Expression-oriented evaluation - Immutable values - Dynamic typing with runtime checks - First-class functions and closures - List comprehensions

Data Types

✅ Implemented: - Scalars: Int, Float, Bool, String - Collections: List, Dict - DataFrames (Arrow-backed) - Vectors (typed arrays) - NA (typed missing values) - Error (structured errors, not exceptions) - Pipeline (DAG execution) - Intent (LLM metadata) - Formula (statistical modeling) - Null

Operators

✅ Implemented: - Arithmetic: +, -, *, /, % - Comparison: ==, !=, <, >, <=, >= - Logical: and, or, not - Pipe: |> (conditional, short-circuits on error) - Maybe-pipe: ?|> (always forwards, including errors) - Formula: ~ (for regression models)

Standard Library

✅ Core Package: - print, pretty_print, type, length, head, tail - map, filter, sum, seq - is_error

✅ Base Package: - error, error_code, error_message, error_context - assert - NA, na_int, na_float, na_bool, na_string, is_na

✅ Math Package: - sqrt, abs, log, exp, pow - min, max

✅ Stats Package: - mean, sd, quantile, cor (with na_rm parameter) - lm (linear regression: y ~ x)

✅ DataFrame Package: - read_csv (with sep, skip_lines, skip_header, clean_colnames) - write_csv - nrow, ncol, colnames - clean_colnames (symbol expansion, diacritics, snake_case, collision resolution)

✅ Colcraft Package (Data Verbs): - select, filter, mutate, arrange - group_by, summarize, ungroup - Window functions: - Ranking: row_number, min_rank, dense_rank, cume_dist, percent_rank, ntile - Offset: lag, lead - Cumulative: cumsum, cummin, cummax, cummean, cumall, cumany

✅ Pipeline Package: - pipeline_nodes, pipeline_deps, pipeline_node, pipeline_run

✅ Explain Package: - explain, explain_json - intent_fields, intent_get

Features

✅ Error Handling: - Errors as values (not exceptions) - Railway-oriented programming with |> and ?|> - Actionable error messages with suggestions

✅ NA Handling: - Explicit NA (typed: na_int(), etc.) - No automatic propagation - na_rm parameter for aggregations - Window functions handle NA gracefully

✅ Pipelines: - DAG-based execution - Automatic dependency resolution - Deterministic order - Cycle detection - Introspection

✅ Intent Blocks: - Structured metadata for LLM collaboration - Document assumptions, constraints, goals - Machine-readable format

✅ Arrow Integration: - Zero-copy columnar storage - CSV reading via Arrow C GLib - Dual-path operations (native + fallback)

✅ Reproducibility: - Nix flakes for dependency management - Deterministic execution - Frozen syntax and semantics

REPL

✅ Implemented: - Interactive read-eval-print loop - Multiline input support - Persistent environment - Auto-loading of standard library

Testing

✅ Implemented: - Unit tests (OCaml) - Golden tests (T vs R comparison) - Example-based tests

Documentation

✅ Implemented (this release): - README with quick start - Getting Started guide - Installation guide - API Reference (all packages) - Language Overview - Data Manipulation Examples - Pipeline Tutorial - Architecture guide - Contributing guide - Development guide - Comprehensive Examples - Error Handling guide - Reproducibility guide - LLM Collaboration guide - FAQ - Troubleshooting guide - Changelog (this file)

Roadmap

Version 0.2.0 (Beta) — Planned

Focus: Performance and expanded functionality

Performance Improvements

• Bytecode compilation (replace tree-walking)
• Stack-based VM
• Lazy evaluation for pipelines
• More Arrow compute kernels

New Features

• DataFrame joins (join, left_join, inner_join)
• More statistical functions (t_test, anova, chisq_test)
• Multiple regression (lm(y ~ x1 + x2 + x3))
• Parquet support
• Basic plotting (via Arrow/matplotlib integration)

Language Enhancements

• Pattern matching on values
• Recursive functions optimization (tail-call)
• Anonymous pipeline blocks
• Destructuring assignment

Standard Library

• String manipulation package (split, join, regex)
• Date/time package (parse_date, diff_days)
• File I/O package (read_lines, write_lines)

Version 0.3.0 (Beta) — Planned

Focus: Ecosystem and tooling

Tooling

• Language Server Protocol (LSP) for IDE support
• Formatter (t-fmt)
• Linter (t-lint)
• Package manager (t-pkg)

Ecosystem

• User-contributed packages
• Package registry
• Binary distribution (no source build)

Language Features

• Optional static typing (gradual typing)
• Type annotations and inference
• Modules and namespaces

Version 1.0.0 (Stable) — Future

Focus: Production-ready, stable API

Stability

• API freeze (no breaking changes)
• Comprehensive benchmarks
• Performance parity with Pandas/dplyr
• 95%+ test coverage

Features

• Distributed execution (multi-node pipelines)
• GPU acceleration (Arrow Compute on CUDA)
• Database integration (SQL, DuckDB)
• Streaming data support

Documentation

• Books and tutorials
• Video courses
• Case studies
• Certification program

Historical Development (Pre-Alpha)

Phase 8: Documentation & Stabilization

• Comprehensive documentation suite
• Website updates
• Alpha freeze

Phase 7: CLI & REPL

• Interactive REPL
• Error recovery
• Standard library auto-loading

Phase 6: LLM Tooling & Introspection

• Intent blocks
• explain() functions
• Pipeline introspection

Phase 5: Math & Statistics

• Math functions
• Statistical functions with NA handling
• Linear regression

Phase 4: Data Verbs & Window Functions

• Six core verbs (select, filter, mutate, arrange, group_by, summarize)
• All window functions (rank, offset, cumulative)
• Grouped operations

Phase 3: Pipelines

• Pipeline syntax
• DAG execution
• Dependency resolution
• Cycle detection

Phase 2: Tabular Data & Arrow

• Arrow C GLib integration
• DataFrame type
• CSV I/O
• Column name cleaning

Phase 1: Language Core

• Lexer and parser
• AST and evaluator
• Basic types and operators
• Functions and closures

Known Issues (Alpha 0.1)

Performance

• Slow for large datasets (>1M rows): Tree-walking interpreter
• Memory usage: No streaming, entire datasets in RAM

Limitations

• No joins: Must pre-join data externally
• No visualization: Export to CSV for plotting
• Limited file formats: CSV only
• No user packages: Fixed standard library

Bugs

• Arrow FFI edge cases: Some Arrow operations may crash on corner cases
• Floating-point precision: Platform-specific differences in golden tests
• Error messages: Some could be more helpful

Reporting: Please report bugs on GitHub Issues

Contributing to Roadmap

We welcome input on priorities! Please: 1. Open GitHub Discussion for feature requests 2. Vote on existing feature requests 3. Contribute implementations (see Contributing Guide)

Versioning Policy

T follows Semantic Versioning (semver):

• Major (1.0.0): Breaking changes
• Minor (0.1.0): New features, backward-compatible
• Patch (0.1.1): Bug fixes

Alpha/Beta status: API may change without major version bump.

Post-1.0: Stable API, breaking changes only in major versions.

Release Schedule

Alpha (current): No fixed schedule, continuous development
Beta: Expected Q3 2026
1.0: Expected 2027

Note: Dates are estimates and may change based on development progress and community feedback.

Stay Updated: Watch the GitHub repository for release announcements!