Skip to contents

Changelog

Source: NEWS.md

resultcheck 0.3.2

  • Built-in defaults now omit broom::augment for fixest, kmeans, drc, factanal, htest, speedlm, speedglm, and glmRob, where augmentation requires extra inputs, only supports particular fits, or always errors. These classes retain tidy and glance; supported augmentation can still be requested explicitly. Existing snapshots containing augmentation will need review.

  • Removed invalid broom defaults for lme, merMod and its subclasses, stanfit, and bamlss. These classes now use the existing print + str fallback unless users configure suitable methods explicitly; no additional dependency is required.

resultcheck 0.3.1

CRAN release: 2026-09-15

  • Snapshot serialization now temporarily disables fancy quotation marks, preventing false differences between interactive R and Quarto model summaries. The caller’s useFancyQuotes option is restored on success and failure.

  • Snapshot serialization now controls max.print locally via snapshot.max_print in _resultcheck.yml (default: 1000 entries), preventing session-dependent truncation of base R printed output. Previously truncated baselines may need to be reviewed and regenerated. Caller options are restored after serialization; class-specific and custom methods can still summarize or explicitly truncate their output.

resultcheck 0.3.0

CRAN release: 2026-09-02

  • snapshot() now detects active Quarto documents across execution engines and active knitr inputs so rendered documents reuse snapshots organized under the document filename instead of "interactive". During Quarto rendering, missing snapshots are created automatically and mismatches stop the render with an error.
  • Snapshot serialization now fixes CLI Unicode formatting so presentation-only differences such as × versus x, … versus ~, and Unicode footnote markers do not vary between interactive sessions and Quarto renders.

resultcheck 0.2.1

CRAN release: 2026-05-08

  • snapshot() now improves script-name detection in non-interactive and test contexts by skipping editor-context lookup there and relying on call-stack source references before falling back to "interactive".

resultcheck 0.2.0

CRAN release: 2026-05-03

  • snapshot() now uses rstudioapi::getSourceEditorContext()$path (in RStudio/Positron) as the primary method to detect the calling script name, falling back to the call stack and then "interactive".
  • snapshot() and serialize_value() now apply built-in class-based method defaults automatically when no explicit method argument is provided. Statistical model classes (e.g. lm, glm, coxph, kmeans) are serialized using broom::tidy, broom::glance, and/or broom::augment when the broom package is available.
  • When broom is not installed, built-in class defaults that rely on it are silently skipped and the print + str fallback is used. User-configured class defaults that reference unavailable packages still raise an error.
  • The full list of supported classes is in inst/extdata/snapshot-method-defaults.R.
  • Added broom to Suggests to enable the new built-in class defaults.
  • Add “Get Started” and FAQ vignettes to clarify workflow and common questions.

resultcheck 0.1.4

CRAN release: 2026-04-21

  • Rename project config file from resultcheck.yml to _resultcheck.yml (legacy resultcheck.yml is still read for backward compatibility).
  • Change default snapshot location to tests/_resultcheck_snaps/ instead of _resultcheck_snapshots/.
  • Add configurable snapshot directory support via snapshot.dir in _resultcheck.yml.
  • Add with_example() to run package examples in temporary projects under tempdir(), avoiding writes to user home/package/getwd locations.
  • Update examples/docs to be executable without \\dontrun{} and use setup_sandbox() with no required file arguments for empty sandboxes.
  • snapshot() now warns with the exact target path before interactive writes; default path remains configurable via snapshot.dir in _resultcheck.yml.

resultcheck 0.1.3

  • Prepare for release to CRAN.

resultcheck 0.1.2

  • Add precision rounding option in _resultcheck.yml to stabilize snapshot comparisons across runs.
  • Add [ignored] markers in snapshot() output to explicitly indicate excluded components.

resultcheck 0.1.1

  • Fixed snapshot inconsistencies caused by variable serialization width and directory checks, improving reliability of result comparisons.
  • Added method parameter to snapshot() for print/str/both serialization.

resultcheck 0.1.0

Initial release

  • snapshot() — create and verify human-readable snapshots of R objects (models, data frames, tables, vectors, etc.). Interactive mode warns and prompts; testing mode (inside run_in_sandbox()) errors on mismatch.
  • setup_sandbox() — copy project files and/or directories into a temporary directory for isolated testing. Accepts individual file paths or entire subdirectories.
  • run_in_sandbox() — execute an R script inside the sandbox with the working directory set to the sandbox root, while find_root() and snapshot() transparently resolve back to the original project root.
  • cleanup_sandbox() — remove the sandbox directory after testing.
  • find_root() — locate the project root using _resultcheck.yml, .Rproj, or .git as markers.