Raise test coverage to 100% (mark unreachable defensive branches)#5
Merged
Conversation
Baseline R-side + compiled coverage was 99.45% (covr 3.6.5, type="tests"), with exactly three uncovered lines, all of which are unreachable defensive branches rather than untested logic: - R/thin.R:77 - the `is.array(image) && length(dim(image)) == 2L` recursion target in as_binary_matrix(). In R (>= 4.2, the package's floor) every object with a length-2 `dim` attribute also satisfies is.matrix(), so the is.matrix() branch above always handles 2-D arrays first. Verified by exhaustively tracing every array/data.frame/list-array input: none reach this line. - src/lee.cpp:42 - `default: on_boundary = 0;` for a `sub` value outside 0-3. `sub` is driven only by the internal `for (int sub = 0; sub < 4; sub++)` loop, so out-of-range values never occur. - src/distance_transform.cpp:235 - `default: Rcpp::stop(...)` for an unknown metric code. The R wrapper validates `metric` with match.arg() and maps it to codes 0/1/2 only, so an out-of-range code never reaches the C++. Per the skip-coverage policy these three (and only these) are wrapped in `# nocov` with a one-line category + reason comment. After the markers, non-nocov coverage is 100% on every R and C++ file. Also adds strict exact-value tests (the existing suite leaned on expect_lt/expect_gt bounds): - tests/testthat/test-coercion-helpers.R (new) - pins as_binary_matrix() foreground-collapse and dimension behaviour, the 2-D-array dispatch equivalence, the documented error messages, and restore_storage() round-trips, all with expect_identical. - exact skeletons for zhang_suen/guo_hall/thinImage in test-thin.R. - exact medial-axis skeleton + Euclidean distance map in test-medial-axis.R. - exact full-matrix manhattan/chessboard/euclidean DTs in test-distance-transform.R. devtools::test(): 0 failures. lintr::lint_package(): 0 lints. R CMD check --as-cran: 0 errors, 0 warnings, 1 NOTE (pre-existing non-portable compilation-flag note on Ubuntu). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Two defensively-unreachable `# nocov` guards previously fell back *silently* when their impossible state was reached. A silent fallback on an "impossible" branch masks the very bug it would signal. Convert both to loud errors that name the violated invariant and the upstream guarantee that makes the branch unreachable, so if the impossible ever happens it is caught rather than hidden. The `# nocov` markers are kept (the branches remain unreachable from R) with their comments relabelled as fail-fast assertions. - R/thin.R as_binary_matrix(): the length-2-`dim` array branch used to silently recurse. In R (>= 4.2) every length-2-`dim` object already satisfies is.matrix() and is handled by the branch above, so this is unreachable; now stop() reports the broken base-R invariant. - src/lee.cpp lee_can_delete(): the switch `default:` used to silently set on_boundary = 0 for a sub-iteration index outside 0-3, but the driving loop is `for (sub = 0; sub < 4; sub++)`; now Rcpp::stop() reports the corrupted loop bound. Kept on one line so the single `// # nocov` excludes the whole statement (matches distance_transform.cpp). src/distance_transform.cpp's `default:` guard was already a loud Rcpp::stop() and is left unchanged. Validation: devtools::test() 0 failures; coverage 100% of non-nocov lines; lintr clean; R CMD check --as-cran 0E/0W/1N (pre-existing Ubuntu compilation-flag NOTE). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Codecov Report✅ All modified and coverable lines are covered by tests.
🚀 New features to boost your workflow:
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Baseline R-side + compiled coverage was 99.45% (covr 3.6.5, type="tests"),
with exactly three uncovered lines, all of which are unreachable defensive
branches rather than untested logic:
is.array(image) && length(dim(image)) == 2Lrecursiontarget in as_binary_matrix(). In R (>= 4.2, the package's floor) every
object with a length-2
dimattribute also satisfies is.matrix(), so theis.matrix() branch above always handles 2-D arrays first. Verified by
exhaustively tracing every array/data.frame/list-array input: none reach
this line.
default: on_boundary = 0;for asubvalue outside 0-3.subis driven only by the internalfor (int sub = 0; sub < 4; sub++)loop, so out-of-range values never occur.
default: Rcpp::stop(...)for an unknownmetric code. The R wrapper validates
metricwith match.arg() and maps itto codes 0/1/2 only, so an out-of-range code never reaches the C++.
Per the skip-coverage policy these three (and only these) are wrapped in
# nocovwith a one-line category + reason comment. After the markers,non-nocov coverage is 100% on every R and C++ file.
Also adds strict exact-value tests (the existing suite leaned on
expect_lt/expect_gt bounds):
foreground-collapse and dimension behaviour, the 2-D-array dispatch
equivalence, the documented error messages, and restore_storage()
round-trips, all with expect_identical.
test-distance-transform.R.
devtools::test(): 0 failures. lintr::lint_package(): 0 lints.
R CMD check --as-cran: 0 errors, 0 warnings, 1 NOTE (pre-existing
non-portable compilation-flag note on Ubuntu).
Co-Authored-By: Claude Opus 4.8 noreply@anthropic.com