aglinxinyuan opened a new pull request, #7957:
URL: https://github.com/apache/texera/pull/7957
### What changes were proposed in this PR?
`dataset-detail.component.spec.ts` goes from 195 tests to 203, closing the
residue left after an earlier pass — mostly the upload-progress paths.
Measured from raw lcov counters, same spec filter on both sides, `rm -rf
coverage` between runs:
| Counter | Before | After |
|---|---|---|
| Lines hit | 412/414 | **414/414 = 100%** |
| Branch arms | 193/203 = 95.1% | **200/203 = 98.5%** (10 missed → 3) |
| Functions | 120/120 | 120/120 |
| Codecov's metric | 411/422 = 97.4% | **420/422 = 99.5%** |
Functions were already 120/120 with zero `FNDA:0` records both before and
after — worth stating, because twice in this campaign a file sat at high line
coverage with functions uncovered and a binding whose handler the spec called
directly stayed `FNDA:0` while looking covered. Not the case here.
### Verification
16 mutations, **all 16 killed**, each named in the mutation table with its
exact failure message.
The first draft reported no survivors. **Seven mutants survived its 202-test
suite with exit 0** — including `.pop()` → `.shift()`, replacing a computed
time-zone name with `""`, and several progress-index substitutions. All now
die, and the search space additionally includes one mutant added on a fresh
axis (the `", "` separator).
Stated with the scope the reviewer correctly insisted on: **this reports the
search space, not a proof about the file.** Sixteen mutants died; that is not
the same claim as "the file is mutation-complete".
Three repairs are worth naming because the original tests looked fine:
- **The time-zone test was degenerate**, asserting on whatever the runner's
ambient `Intl` formatter produced. It now stubs the formatter, so the assertion
is about the component's own `.split(", ").pop()` parsing rather than the
platform's output — and a second test covers the separator.
- **Two single-index tests passed trivially.** "Ignores a hide request for a
row that is gone" and the basename test each now assert a valid-index half
alongside the invalid one, so an index substitution cannot slip through.
- One reported failure mode was simply wrong and is rewritten: a mutant was
described as surfacing a `TypeError` through a `.not.toThrow()` assertion, when
the emission actually carries a valid percentage.
### Deliberately not included
Two lines with three branch arms remain, and both are refused for the same
reason: `percentage: progress.percentage ??
this.uploadTasks[taskIndex].percentage ?? 0` (line 647) and its twin in the
error handler (line 678) fall through a `??` whose right operand no caller can
produce. Verified in the final lcov — line 647 is hit 28 times with both arms
at zero.
After this bundle those are the **only** two lines in the file still
carrying a missed arm, and the zero-hit set is empty.
Also recorded honestly: **three** of the tests pin defensive paths
production cannot reach (the first draft said two). They are kept because they
document the guards, not because they earn coverage.
No production file is touched, and the `node_modules` junction used for the
run was removed before committing.
### Any related issues, documentation, discussions?
Closes #7955
### How was this PR tested?
```
npx ng test --watch=false --include="**/dataset-detail.component.spec.ts"
```
```
Test Files 1 passed (1)
```
`yarn format:ci` passes. `frontend/junit.xml` and `frontend/coverage/` are
regenerated by every run and are not committed.
### Was this PR authored or co-authored using generative AI tooling?
Generated-by: Claude Code (Opus 5)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
