| #!/usr/bin/env python3 | |
| """ | |
| calculate_counterfactuals_v7_master.py | |
| 100% Full-Fidelity Counterfactual Replay of `po/` and `src/library/*/po` Commit History. | |
| Implements all four final reproducibility recommendations to achieve ultimate speed, safety, and exactitude: | |
| 1. Eliminated 20-Minute AST Bottleneck (`blob_ast_cache[sha]`): Parses each of the 24,171 unique historical | |
| blobs exactly ONCE into `blob_ast_cache[sha]` during batch loading. Total AST parsing time across all | |
| 4 scenarios (`194,000` evaluations) drops from ~15 minutes down to ~3 seconds. | |
| 2. RAMdisk Inode/Memory Protection (1,000-Pair Batches): Processes modifications in batches of `1,000 commit pairs` |
| library(RSQLite) | |
| git_repo <- "~/git/r-svn" | |
| svn_repo <- "~/svn/r-devel" | |
| if (!dir.exists(file.path(git_repo, ".git"))) { | |
| stop(sprintf("Error: Git repository not found at '%s'. Please provide a valid git repo path as arg 1.", git_repo)) | |
| } | |
| Sys.setlocale("LC_ALL", "C") |
| library(bench) | |
| library(dplyr) | |
| library(tidyr) | |
| run_benchmark_pair <- function(N, type, uniqueness, na_ratio = 0, nan_ratio = 0) { | |
| cat(sprintf("Running: N=%d, type=%s, uniqueness=%s, na=%f, nan=%f\n", N, type, uniqueness, na_ratio, nan_ratio)) | |
| # Generate data | |
| if (uniqueness == "high") { | |
| num_unique <- N |
Provide a secondary audit of the performance claims regarding sort(unique(x)) vs unique(sort(x)).
We compare these two expressions across various input sizes, data types, and presence/absence of missing values (NA / NaN).
Benchmarks were performed using the bench package in R.
We tested the following parameters:
| # nolint start: line_length_linter, object_name_linter. | |
| library(methods) | |
| # Setup paths | |
| myLib <- normalizePath("myLib") | |
| # Read scan results to get all unique matched packages | |
| results_data <- readRDS("s4_scan_results.rds") | |
| all_pkgs <- unique(sapply(strsplit(sapply(results_data, `[[`, "package"), "_"), `[`, 1)) |
| # This script creates two dummy packages to prove that the binary search evaluation | |
| # correctly identifies packages that fail when digits.rank is set too low. | |
| # -- Create dummy3 -- | |
| # This package expects wilcox.test(c(1.11, 2.22), c(1.14, 2.25))$statistic to be exactly 1. | |
| # This assertion fails at digits.rank=1 and 2, but passes at digits.rank >= 3. | |
| dir.create("dummy3/R", recursive = TRUE, showWarnings = FALSE) | |
| dir.create("dummy3/tests/testthat", recursive = TRUE, showWarnings = FALSE) | |
| writeLines(c( |
| # Assume these are installed in a state to pass R CMD check, i.e., | |
| # with enough Suggests to pass with _R_CHECK_FORCE_SUGGESTS_=false | |
| packages <- c("cardx", "caTools", "clintools", "cogmapr", "CRMetrics", "EasyDescribe", | |
| "effectsize", "EnvStats", "eyetrackingR", "ggpubr", "ggpval", "ggsignif", | |
| "gtsummary", "iCellR", "iDOS", "jsmodule", "LGDtoolkit", "microeco", | |
| "mnda", "mt", "PairedData", "pairwiseCI", "papeR", "pctax", "pcutils", | |
| "PLEXI", "plotbb", "plotthis", "PopComm", "qPCRtools", "RadOnc", | |
| "rattle", "rbiom", "Rcmdr", "RcmdrPlugin.MPAStats", "rcompanion", | |
| "rempsyc", "ReporterScore", "SCpubr", "sigminer", "tinyarray", | |
| "TOSTER", "UCSCXenaShiny", "voiceR", "volcano3D") |
| #!/bin/bash | |
| set -e # Exit immediately if a command exits with a non-zero status. | |
| # --- 1. Environment Setup --- | |
| echo ">>> Installing System Dependencies..." | |
| sudo apt-get update -qq | |
| # Dependencies: | |
| # - flex/bison: for udunits parser compilation | |
| # - texinfo: for udunits documentation | |
| # - libexpat1-dev: for udunits XML parsing |
| cran_repo = "https://cloud.r-project.org" | |
| bioc_repo = BiocManager::repositories()["BioCsoft"] | |
| cran_db <- data.frame(available.packages(repos = cran_repo)) | |
| bioc_db <- data.frame(available.packages(repos = bioc_repo)) | |
| methods_importers <- function(db, skip) db |> | |
| subset( | |
| grepl("(^|[^\\w.])methods($|[^\\w.])", Imports, perl=TRUE), | |
| "Package", |
| db <- data.frame(available.packages()) | |
| methods_imports <- db |> | |
| subset(grepl("(^|[^\\w.])methods($|[^\\w.])", Imports, perl=TRUE), "Package", drop=TRUE) | |
| test_lib = '/media/WesternDigital3839/tmpCRAN' | |
| # very slow, and requires some iteration to get SystemRequirements | |
| install.packages(methods_imports, lib=test_lib) |