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

stillwater-sc / universal / 43564593067
89%
master: 84%

Build:
Build:
LAST BUILD BRANCH: fix/fixpnt-quire-resolve-saturates
DEFAULT BRANCH: master
Ran 14 Nov 2026 06:26AM UTC
Jobs 1
Files 890
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

14 Nov 2026 06:05AM UTC coverage: 89.062% (+0.01%) from 89.052%
43564593067

push

github

web-flow
fix(native): find_msb(long) on LP64, NaN payloads, ulp and fastipow edges, setbit bounds, signed-zero universal_cast (#2063)

* fix(native): find_msb(long) on LP64, NaN payload classification, ulp and fastipow edges, setbit bounds, signed-zero universal_cast

The native and utility findings of QA sweep #2039, except the cross-type
promotion rule, which waits on a decision (#2059 stays open for it).

- find_msb(long) scanned 32 bits: find_msb(1L << 40) was 0 on LP64, where
  int64_t is long. It dispatches on sizeof(long).
- checkNaN recognised only the canonical patterns: every other payload
  (0x7f800001, 0x7fc00001, 0x7fffffff, 0x7ff0000000000001) was not NaN. A NaN
  is now any all-ones exponent with a nonzero fraction, quiet by the quiet bit
  (MSVC keeps its signaling_NaN() pattern as signalling); long double gets
  checkNaN/checkInf of its own (x87, binary128, double and IBM layouts).
- ulp() is the magnitude of the spacing, finite at max() (the gap below) and inf
  at inf; it was negative for negative arguments, inf at max() and NaN at inf.
  Its one caller, the cfloat conversion suite, applies the sign itself.
- fastipow computes in the unsigned domain with overflow checks: 3^41 was
  signed-overflow UB, (-2)^63 = INT64_MIN came out 0. Overflow still returns 0.
- setbit(float&/double&) ignores an index past the width (was a UB shift).
- to_binary/to_hex go bit by bit through the unsigned type: nbits > 64 is the
  sign extension instead of a shift past 64 bits.
- universal_cast/universal_assign carry the sign of a zero (Target{} was +0).
  The <= 52 threshold is right: precision_bits counts fraction bits.

Tests (fail on main): static/native/float/special_patterns.cpp,
static/native/integer/edges.cpp.

Relates to #2059

BEHAVIOUR-CHANGE: native ulp(x) is a positive magnitude for negative x, finite at max(); checkNaN classifies every NaN payload; find_msb(long) is correct on LP64; fastipow((-2), 63) is INT64_MIN and overflow is never UB; universa... (continued)

63 of 63 new or added lines in 6 files covered. (100.0%)

2 existing lines in 1 file now uncovered.

62931 of 70660 relevant lines covered (89.06%)

11050523.72 hits per line

Coverage Regressions

Lines Coverage ∆ File
2
97.67
-0.13% include/sw/universal/number/cfloat/cfloat_impl.hpp
Jobs
ID Job ID Ran Files Coverage
1 43564593067.1 14 Nov 2026 06:26AM UTC 890
89.06
GitHub Action Run
Source Files on build 43564593067
  • Tree
  • List 890
  • Changed 8
  • Source Changed 7
  • Coverage Changed 8
Coverage ∆ File Lines Relevant Covered Missed Hits/Line
  • Back to Repo
  • Github Actions Build #43564593067
  • 5576740d on github
  • Prev Build on main (#43559744627)
  • Next Build on main (#43564838215)
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