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

stillwater-sc / universal / 42155020823
89%
master: 84%

Build:
Build:
LAST BUILD BRANCH: fix/fixpnt-quire-resolve-saturates
DEFAULT BRANCH: master
Ran 06 Nov 2026 03:27AM UTC
Jobs 1
Files 848
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

06 Nov 2026 03:11AM UTC coverage: 88.603% (-0.008%) from 88.611%
42155020823

push

github

web-flow
fix(numeric): faithful's numeric_limits, extremes and attributes.hpp (#1951)

std::numeric_limits<faithful<Real>> was a stub. min(), max() and
lowest() were zero, epsilon() and denorm_min() one, and digits 3333333.
The free minpos, maxpos, minneg and maxneg returned their argument
unchanged. attributes.hpp was another type's, written for nbits and a
block type, and the umbrella left it out because it did not compile.

A faithful<Real> is a two-part expansion like dd, so its limits follow
the convention of the multi-double types (#1882), with Real in place of
double:
- min() is the full-precision floor, 2^(min_exponent - 1), with
  min_exponent = Real's + Real's digits. That is 2^-969 for double, as for
  dd, and 2^-102 for float.
- epsilon() is eps(Real)^2 / 2: 2^-105 for double, 2^-47 for float.
- max() is Real's maximum plus max * 2^-(digits + 1), the largest error
  that does not round the pair up to infinity. lowest() is -max().
- denorm_min() is Real's subnormal minimum. infinity() and the NaNs are
  (x, 0). digits is twice Real's, and max_digits10 is
  ceil(1 + digits * log10 2).
- has_infinity, the NaN flags and has_denorm are Real's. round_style is
  round_to_nearest and is_bounded true, as for dd.

minpos and maxneg are the extremes: (denorm_min, 0) and -max(). The
limits are constant expressions, as for the native types. The pair
constructor, unary minus and high()/low() are constexpr, and a constexpr
pow2 computes the powers of two, since std::ldexp is not constexpr.

attributes.hpp now has sign(), scale() and faithful_range(), which lists
maxneg ... minneg 0 minpos ... maxpos. The umbrella includes it.

numeric/faithful/limits_and_attributes.cpp checks every limit for float
and double, the four extremes, and the attributes. The dd-equal values
of faithful<double> are checked with static_assert.

Resolves #1949

BEHAVIOUR-CHANGE: std::numeric_limits<faithful<Real>> now returns real limits instead of zeros and ones (min, max, lowest, epsilon... (continued)

128 of 148 new or added lines in 4 files covered. (86.49%)

4 existing lines in 3 files now uncovered.

57738 of 65165 relevant lines covered (88.6%)

8851463.45 hits per line

Uncovered Changes

Lines Coverage ∆ File
20
79.59
numeric/faithful/limits_and_attributes.cpp

Coverage Regressions

Lines Coverage ∆ File
2
84.93
-0.3% include/sw/universal/number/qd/qd_impl.hpp
1
97.8
-0.06% include/sw/universal/number/cfloat/cfloat_impl.hpp
1
93.16
0.0% include/sw/universal/verification/correctly_rounded_reference.hpp
Jobs
ID Job ID Ran Files Coverage
1 42155020823.1 06 Nov 2026 03:27AM UTC 848
88.6
GitHub Action Run
Source Files on build 42155020823
  • Tree
  • List 848
  • Changed 4
  • Source Changed 1
  • Coverage Changed 4
Coverage ∆ File Lines Relevant Covered Missed Hits/Line
  • Back to Repo
  • Github Actions Build #42155020823
  • 3ff06362 on github
  • Prev Build on main (#42150180977)
  • Next Build on main (#42161765292)
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