Changelog

randomwalk (development version)

Changes

• Archived dynamic_broadcasting.qmd vignette - consolidated into dashboard_comprehensive.qmd which has all the same features plus better organization.

• Simplified dashboard_comprehensive.qmd - moved documentation into Notes tab, removed redundant sections.

• Updated telemetry vignette to handle missing coverage data gracefully.

randomwalk 2.1.0

Breaking Changes

• sync_mode = "dynamic" and sync_mode = "mirai_dynamic" are now DEPRECATED. These modes silently fall back to static behavior because nanonext sockets cannot be created inside crew/mirai subprocesses (NNG fork boundary limitation). Use sync_mode = "chunked" instead for best collision detection (~15%).

• Removed unused duckdb dependency from DESCRIPTION and all Nix files.

New Features

• Added sync_mode = "chunked" - processes walkers in batches of 10 with grid updates between batches. This is now the recommended mode for parallel simulations with collision detection.

• New shell.nix for end users - uses builtins.fetchGit to install randomwalk directly from GitHub without requiring sha256 hashes.

• Added R-universe badge to README for package availability status.

Bug Fixes

• Fixed plot_grid() collision count display - now correctly checks collisions_detected field used by async modes, with fallback to multiple termination reason key names.

• Fixed S3 plot() method signature to use x and ... to match the generic.

• Fixed validation to properly respect boundary parameter using match.arg().

• Fixed chunked mode to use numeric 1 instead of string for black pixels.

Documentation

• Updated README sync_mode table with deprecation warnings for dynamic modes.

• Updated dynamic_broadcasting_algorithm.md with explanation of why the socket-based approach failed.

• Updated pkgdown home page title and navigation labels.

• Fixed R-CMD-check badge URL (lowercase filename).

randomwalk 2.0.0

Major Changes

• Complete rewrite with async parallel processing support via crew backend.

• Added comprehensive statistics tracking with percentiles and formatting.

• Added S3 plot methods - plot(result) works directly.

• Added WebR/browser support via Shinylive dashboards.

• Added multiple sync modes: static, chunked, dynamic, mirai_dynamic.

New Features

• run_simulation() - main entry point with full parameter control.

• plot_grid() - visualize final grid state with statistics.

• plot_walker_paths() - visualize walker trajectories.

• run_dashboard() - launch interactive Shiny dashboard (native R only).

• format_statistics() - format simulation results for display.

Documentation

• Added comprehensive vignettes: defensive programming, step distribution analysis, telemetry.

• Added browser-based Shinylive dashboards for interactive demos.

• Added pkgdown site at https://johngavin.github.io/randomwalk/

randomwalk 1.0.0

• Initial release with basic random walk simulation.