Reporting bugs - #159
Merged
Merged
Conversation
When the container is a string, containsElement rendered the element with reflect.Value.String(). That returns a placeholder for any kind other than a string, so Contains(t, "hello", 'e') compared "hello" against the literal "<int32 Value>" and failed, naming a type where the character should be. stringContainsElement now dispatches on the element instead: strings.Contains for a string, strings.ContainsRune for a rune, strings.IndexByte for a byte, and the reflect fallback only for a defined string type such as `type Doc string`. Any other kind cannot occur in a string and still reports "does not contain". Left alone: a needle that is not valid UTF-8 still matches on byte boundaries, so Contains(t, "é", "\xa9") stays true, since "\xa9" is the second byte of "é". Upstream converts both operands to []rune to prevent that, which costs an allocation on every string Contains to reject a deliberately malformed needle. reference: github.com/stretchr#1908 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Signed-off-by: Frederic BIDON <fredbi@yahoo.com>
Empty rendered the value with %v, so a string of whitespace left nothing on screen: Empty(t, " ") reported `Should be empty, but was `, which reads as a message that lost its value. NotEmpty had the same problem with the empty string. truncatingValue picks %q for anything of string kind, aliased string types included, and keeps %v for everything else. NotEmpty now goes through it too, so its value is truncated at maxMessageSize like every other failure message instead of being printed whole. reference: github.com/stretchr#1874 reference: github.com/stretchr#1875 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Signed-off-by: Frederic BIDON <fredbi@yahoo.com>
The per-element call passed "at index %d" as the message, which does not add the index to what the caller wrote, it replaces it. A caller who passed InEpsilonSlice(t, want, got, eps, "checking %s", "widths") saw the index and never saw "checking widths". withContext joins the two: the positional context, then the caller's message rendered through messageFromMsgAndArgs. With nothing from the caller it yields the context alone, so a bare call still names the element that failed. reference: github.com/stretchr#1931 reference: github.com/stretchr#1899 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Signed-off-by: Frederic BIDON <fredbi@yahoo.com>
A missing key was already reported by name, but a key present in both maps whose values are further apart than delta was not: the failure came from InDelta, which sees the two values and not the key they were read under. On a map of any size that leaves the reader to work out which entry failed. The per-key call now carries "at key <k>" through withContext, the same way InEpsilonSlice carries its index, so the caller's own message survives too. reference: github.com/stretchr#1898 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Signed-off-by: Frederic BIDON <fredbi@yahoo.com>
The per-element call handed InDelta the actual value first and the expected
one second, so a failure read "Max difference between 9 and 2" for
InDeltaSlice(t, []float64{2}, []float64{9}, 0.5) — the two named the wrong way
round. The verdict was never affected, since a delta is symmetric, which is why
only a message test catches it.
Inherited from upstream, where the same two arguments are still swapped. No
upstream issue covers it.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Frederic BIDON <fredbi@yahoo.com>
EqualValues(t, int64(-1), uint64(math.MaxUint64)) passed. ObjectsAreEqualValues compares by converting one operand to the other's type, and converting a negative signed integer to an unsigned one wraps it to exactly the value it is being compared against. Converting the smaller type to the larger one, which is what keeps EqualValues(t, int(270), int8(14)) false, does not help here: int8(-1) wraps to uint64(math.MaxUint64) just as int64(-1) does. Four shapes passed wrongly — int64/uint64, int8/uint64, int/uint, and either argument order. A signed and an unsigned integer now compare by value instead: a negative signed value equals no unsigned value, and otherwise both fit in a uint64. Floats are untouched, and mixed signedness still compares equal when the values are, as in EqualValues(t, uint8(200), int64(200)). Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Signed-off-by: Frederic BIDON <fredbi@yahoo.com>
The August sweep of stretchr/testify turned up four reports describing bugs we had inherited. Each now has a row in the implemented table saying what we adopted and what we left alone, and stretchr#1942 joins stretchr#1940 as the PR behind ErrorNotContains. stretchr#1776 goes in as informational: upstream's generated require documentation lists a bool return that the functions do not have, the same defect our own API pages carried until the signature tables were corrected. Nothing to adopt, but worth recording that both projects found it independently. Two more bugs surfaced while checking those four, neither of them reported upstream, so neither gets a row: InDeltaSlice named expected and actual the wrong way round in its failures, and EqualValues compared a negative signed integer equal to a large unsigned one. Also in this file: the review-frequency line was three months stale, [stretchr#1937] was referenced with no link definition and rendered as literal text, three definitions were duplicated, and the summary counts are recounted from the tables. reference: github.com/stretchr#1776 reference: github.com/stretchr#1874 reference: github.com/stretchr#1875 reference: github.com/stretchr#1898 reference: github.com/stretchr#1899 reference: github.com/stretchr#1908 reference: github.com/stretchr#1931 reference: github.com/stretchr#1940 reference: github.com/stretchr#1942 Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Signed-off-by: Frederic BIDON <fredbi@yahoo.com>
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## master #159 +/- ##
==========================================
+ Coverage 93.68% 93.71% +0.02%
==========================================
Files 106 106
Lines 13842 13883 +41
==========================================
+ Hits 12968 13010 +42
+ Misses 867 866 -1
Partials 7 7 ☔ View full report in Codecov by Harness. |
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.
Change type
Please select: 🆕 New feature or enhancement|🔧 Bug fix'|📃 Documentation update
Short description
Fixes
Full description
Checklist