Fix getRows/getCols filtering and reporting; bump version; add audit

[vertesy]

Motivation

• Fix incorrect and failing behavior in row/column selection helpers where remove0only = TRUE referenced an undefined mx and could error in clean sessions.
• Correct reporting of missing requested names which used the set-difference in the wrong direction and did not reflect the requested-but-missing items.
• Add compact input assertions and record an audit to document findings and deferred items before further, larger changes.

Description

• Replace mx usage in getRows() and getCols() with localized selected_mat <- mat[... , drop = FALSE] and filter rows/columns from that selection to avoid relying on an external object.
• Add compact stopifnot(...) input assertions to getRows() and getCols() to validate key arguments and preserve dimensions by returning with drop = FALSE.
• Bump package version from 2.8.0 to 2.8.1 in Development/config.R and DESCRIPTION and add Development/Audit_20260827.md with a checklist of audit findings and deferred fixes.

Testing

• Static regression assertions (Python checks that scan for the bad patterns and version updates) passed.
• Structural check git diff --check (pre-commit style) reported no problems.
• R CMD check and lintr::lint_package() were attempted but could not run in this environment because R / Rscript are not installed, so package checks and linter runs are pending in an R-equipped environment.

---

Codex Task

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: bce0609de3

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Allow sparse matrix classes in the input check

Problem: getRows() now rejects S4 matrix implementations such as Matrix::dgCMatrix; the identical assertion in getCols() does the same. These objects are not base matrices or data frames, although the existing name lookup, subsetting, and row/column summary operations support them. Trigger: Any downstream caller passes a sparse matrix, including common Seurat workflows. Fix: Validate the required two-dimensional and named interface, or explicitly accept compatible Matrix classes, instead of restricting mat to base matrices and data frames.

AGENTS.md reference: AGENTS.md:L46-L50

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Count matched names before applying value filters

Problem: length(idx) counts rows retained after zero/NA filtering, but the message describes that number as rows that “are found.” Trigger: If a requested row exists but is removed by remove0only or removeNAonly, getRows() can print 0 / 1 are found. Missing: 0; getCols() has the same issue. Fix: Capture the matched-name count immediately after intersect(), or report the matched and retained counts separately.

Useful? React with 👍 / 👎.

[vertesy]

@copilot

Please validate and resolve problems:

1
"Allow sparse matrix classes in the input check
Problem: getRows() now rejects S4 matrix implementations such as Matrix::dgCMatrix; the identical assertion in getCols() does the same. These objects are not base matrices or data frames, although the existing name lookup, subsetting, and row/column summary operations support them. Trigger: Any downstream caller passes a sparse matrix, including common Seurat workflows. Fix: Validate the required two-dimensional and named interface, or explicitly accept compatible Matrix classes, instead of restricting mat to base matrices and data frames."

2
"P2 Badge Count matched names before applying value filters
Problem: length(idx) counts rows retained after zero/NA filtering, but the message describes that number as rows that “are found.” Trigger: If a requested row exists but is removed by remove0only or removeNAonly, getRows() can print 0 / 1 are found. Missing: 0; getCols() has the same issue. Fix: Capture the matched-name count immediately after intersect(), or report the matched and retained counts separately."

Thank you!

[Copilot]

@copilot

Please validate and resolve problems:
...

Addressed in fda93f8.

• Updated getRows() / getCols() input checks to also accept Matrix-compatible S4 inputs (methods::is(mat, "Matrix")) while keeping required row/column name checks.
• Fixed badge/reporting semantics by capturing matched-name count immediately after intersect() and reporting both matched and retained counts after optional NA/zero filters.