stillwater-sc / universal / 26489993803
84%
master: 84%

LAST BUILD BRANCH: main
DEFAULT BRANCH: master
Ran
Jobs 1
Files 661
Run time 2min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 84.003% (+0.02%) from 83.982%
26489993803

push

github

web-flow 
test(ereal): add parse-cost benchmark regression target (#1013) (#1016)

* test(ereal): add parse-cost benchmark regression target (#1013)

Follow-up to #913 (closed): the ereal::parse O(N^3) complexity blowup is fixed
(expansion_product renormalize #981 + maxlimbs digit/result caps in #1006/#1011),
but #913's acceptance asked for a benchmark target tracking parse cost vs string
length. Adds benchmark/performance/arithmetic/ereal/parse.cpp measuring
ereal<2|8|19>::parse over 32..440-digit strings, wired into the arithmetic
benchmark CMakeLists (GLOB + compile_all, mirroring the sibling types).

It guards against a reintroduction of the blowup with a deliberately enormous
threshold (2s for a 320-digit parse) so it never flakes on a slow/loaded CI
runner -- only the catastrophic O(N^3) regression (was >120s, ~5ms today) trips
it, not normal timing variation.

Measured (clang -O3): 320-digit parse = 0.20ms (ereal<2>) / 5.5ms (ereal<19>).
Builds and runs on gcc+clang.

Resolves #1013

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* test(ereal): address PR #1016 review - check parse() return, fix mean wording

CodeRabbit review on the parse-cost benchmark:
- time_parse_ms ignored parse()'s bool return; a failed parse (returns false,
  leaves the value unchanged) would have been timed as valid work. It now counts
  failures into parseFailures, and the benchmark fails if any occur (every input
  is a valid decimal, so the count must be 0).
- the timing comment said "median" but the function returns total/reps (the
  arithmetic mean); reworded to "average".

Verified gcc+clang.

Relates to #1013

---------

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

46878 of 55805 relevant lines covered (84.0%)

5774218.34 hits per line

Jobs
ID Job ID Ran Files Coverage
1 26489993803.1 661
84.0
 GitHub Action Run
Source Files on build 26489993803