|
Ran
|
Jobs
1
|
Files
519
|
Run time
2min
|
Badge
README BADGES
|
push
github
perf: fix `perf/tuple.cc` benchmark The commit fixes several tuple benchmark problems (and potential unwanted optimization is among them): 1. There was `std::vector<char[...]>` construction - it looks monstrous and actually breaks build on some platforms (MacOS, for example) because of destructor problems. Let's replace it with `std::array`, it has the same semantics and is actually the right way to go because it is a C++ class and has a defined destructor. 2. There is an annoying warning complaining that `sprintf` is deprecated, so let's simply replace it with `snprintf`. 3. There are warnings regarding `DoNotOptimize` usage - its const-reference overload is deprecated because it can be actually optimized by compiler. Let's save expression results to local variables so that non-const reference overload is used. NO_TEST=perf NO_CHANGELOG=perf NO_DOC=perf (cherry picked from commit a854a6519)
69106 of 116467 branches covered (59.34%)
103939 of 118684 relevant lines covered (87.58%)
1373615.31 hits per line
| Lines | Coverage | ∆ | File |
|---|---|---|---|
| 1 |
95.96 |
-0.18% | src/box/relay.cc |
| 1 |
80.74 |
-0.33% | src/box/tuple_bloom.c |
| 1 |
90.9 |
-0.1% | src/box/vy_scheduler.c |
| 1 |
80.0 |
-2.05% | src/lib/core/clock.c |
| 1 |
92.83 |
-0.1% | src/lib/core/fiber.c |
| 1 |
96.55 |
-1.15% | src/lib/core/histogram.c |
| 1 |
91.56 |
-0.11% | src/lib/swim/swim.c |
| 2 |
94.25 |
-0.29% | src/box/memtx_space.c |
| 2 |
87.5 |
-3.41% | src/box/xlog.h |
| 2 |
85.61 |
-0.72% | src/lib/core/coio.c |
| 3 |
84.43 |
-0.03% | src/box/xlog.c |
| 4 |
87.44 |
-0.12% | src/box/vy_run.c |
| ID | Job ID | Ran | Files | Coverage | |
|---|---|---|---|---|---|
| 1 | 18749376016.1 | 519 |
87.58 |
GitHub Action Run |
| Coverage | ∆ | File | Lines | Relevant | Covered | Missed | Hits/Line | Branch Hits | Branch Misses |
|---|