• Home
  • Features
  • Pricing
  • Docs
  • Announcements
  • Sign In

stillwater-sc / universal / 30681506242
85%
master: 84%

Build:
Build:
LAST BUILD BRANCH: main
DEFAULT BRANCH: master
Ran 01 Aug 2026 03:50AM UTC
Jobs 1
Files 600
Run time 1min
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

01 Aug 2026 03:11AM UTC coverage: 85.234% (+0.004%) from 85.23%
30681506242

push

github

web-flow
fix(cfloat): clear -Wsign-conversion in cfloat + blocktriple (#1282) (#1287)

* fix(cfloat/blocktriple): clear -Wsign-conversion (#1282)

Sub-issue of Epic #1279. Six sites, all value-preserving:

blocktriple.hpp roundingDecision (299-300): `shift + adjustment` mixes
unsigned `shift` with int `adjustment` (subnormal adjustment). adjustment
was already implicitly converted to unsigned; make it explicit
(`shift + static_cast<unsigned>(adjustment)`) -- bit-identical.

cfloat_impl.hpp:
- setsign (1268): `~SIGN_BIT_MASK` promotes the narrow bt mask to int
  (value-changing for uint8/uint16 limbs); cast the complement back to bt
  before the &= -- the retained low bits are identical.
- convert_signed_integer (2543): `sizeInBits - exponent - 1` with int
  exponent (msb of a nonzero value, so 0 <= exponent < sizeInBits) -> cast
  exponent to uint32_t; exact since non-negative and in range.
- subnormal convert (2940): the sticky-bit mask shift count is now computed
  in signed int (`static_cast<int>(fbits) + exponent + ...`), matching the
  sibling areal code; exponent < 0 on this path, and the sum is a valid
  small shift count. Same value, no unsigned wrap of the int operands.
- to_string(value, precision) (3534): `lsbScale = scale - fbits` is negative
  whenever |scale| < fbits, and a raw size_t cast would spin powerOf2 ~2^64
  times. Guard the conversion (clamp negative to 0) before casting -- fixes
  a latent hang as well as the warning. (This legacy overload is superseded
  by the blocktriple-based operator<< and is already lossy for fractional
  values; out of scope to rewrite here.)

Verified -Wsign-conversion clean on gcc AND clang (no new -Wconversion);
blocktriple rounding/addition/api, cfloat integer_conversion/
ieee754_subnormals/assignment, and cfloat subnormal addition+multiplication
(REGRESSION_LEVEL_1) PASS on both; guarded to_string no longer hangs on
fractional inputs.

Relates to #1279
Resolves #1282

Co-Authored-By: Claude Opus 4.8 (1M con... (continued)

4 of 5 new or added lines in 2 files covered. (80.0%)

8 existing lines in 1 file now uncovered.

41710 of 48936 relevant lines covered (85.23%)

7271778.77 hits per line

Uncovered Changes

Lines Coverage ∆ File
1
93.73
-0.47% include/sw/universal/number/cfloat/cfloat_impl.hpp

Coverage Regressions

Lines Coverage ∆ File
8
93.73
-0.47% include/sw/universal/number/cfloat/cfloat_impl.hpp
Jobs
ID Job ID Ran Files Coverage
1 30681506242.1 01 Aug 2026 03:50AM UTC 600
85.23
GitHub Action Run
Source Files on build 30681506242
  • Tree
  • List 600
  • Changed 3
  • Source Changed 2
  • Coverage Changed 2
Coverage ∆ File Lines Relevant Covered Missed Hits/Line
  • Back to Repo
  • Github Actions Build #30681506242
  • 5ef36871 on github
  • Prev Build on main (#30672136635)
  • Next Build on main (#30683503007)
STATUS · Troubleshooting · Open an Issue · Sales · Support · CAREERS · ENTERPRISE · START FREE TRIAL · SCHEDULE DEMO
ANNOUNCEMENTS · TWITTER · TOS & SLA · Supported CI Services · What's a CI service? · Automated Testing

© 2026 Coveralls, Inc