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

stillwater-sc / universal / 37328409057
86%
master: 84%

Build:
Build:
LAST BUILD BRANCH: main
DEFAULT BRANCH: master
Ran 05 Oct 2026 03:22PM UTC
Jobs 1
Files 684
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

05 Oct 2026 02:53PM UTC coverage: 86.144% (+0.05%) from 86.097%
37328409057

push

github

web-flow
feat(bposit): the bounded posit, Gustafson's b-posit <N, rS, eS> (#1634)

* feat(bposit): the bounded posit, Gustafson's b-posit <N, rS, eS>

A posit whose regime is capped at rS bits (Gustafson, "Every Bit Counts";
arXiv 2603.01615). The regime is a run that counts its terminating bit, and
a run that reaches rs ends without one, so it takes 2..rs bits, r lies in
[-rs, rs-1], and the exponent is always the full es bits. Every encoding
keeps at least nbits - 1 - rs - es fraction bits; the scale is bounded to
[-rs 2^es, rs 2^es - 1] -- 2^-192 .. 2^192 for the standard rS = 6, eS = 5.

Spec: docs/number-systems/bposit.md (#1251), with the range/precision
profile, numeric_limits, a reference encoding table for bposit<8,4,0>, and
the negative result that bounding the taper does NOT make TwoSum or
TwoProduct exact (exhaustive counts for two 8-bit configurations, and
failures in bposit<16,6,5>).

Type: bposit<nbits, rs, es, bt>, with bposit16/32/64 = bposit<n, 6, 5>.
- Compile-time constraints: 2 <= rs < nbits - 1, nbits > 1 + rs + es (a
  fraction bit at every magnitude), nbits <= 64.
- Posit-standard rounding: nearest, ties to the even encoding; never to 0
  or NaR; clamped to +/-minpos and +/-maxpos. 2^minscale exactly is the
  zero pattern and clamps to minpos.
- A 64-bit integer codec (constexpr); + - * / through blocktriple,
  rounded once; native conversions decompose the double exactly;
  comparison is integer order with NaR below every real.
- Trivially constructible; numeric_limits, type_tag, to_binary
  (sign.regime.exponent.fraction), bposit_range, iostream; layered
  core / manipulators / iostream / attributes headers (#1334).
- UNIVERSAL_BUILD_NUMBER_BPOSITS, in the statics cascade.

Tests (static/tapered/bposit):
- conversion: every encoding of six small configurations and both 16-bit
  ones against an independent bit-string decoder -- value, monotonicity,
  F_min fraction bits, maxpos/minpos formulas, double round trip, every
  midpoint ties-... (continued)

261 of 265 new or added lines in 5 files covered. (98.49%)

9 existing lines in 1 file now uncovered.

46359 of 53816 relevant lines covered (86.14%)

9818809.85 hits per line

Uncovered Changes

Lines Coverage ∆ File
2
66.67
include/sw/universal/number/bposit/exceptions.hpp
2
94.12
include/sw/universal/number/bposit/manipulators.hpp

Coverage Regressions

Lines Coverage ∆ File
9
95.35
-0.52% include/sw/universal/number/cfloat/cfloat_impl.hpp
Jobs
ID Job ID Ran Files Coverage
1 37328409057.1 05 Oct 2026 03:22PM UTC 684
86.14
GitHub Action Run
Source Files on build 37328409057
  • Tree
  • List 684
  • Changed 1
  • Source Changed 0
  • Coverage Changed 1
Coverage ∆ File Lines Relevant Covered Missed Hits/Line
  • Back to Repo
  • Github Actions Build #37328409057
  • 6aaee570 on github
  • Prev Build on main (#37160643253)
  • Next Build on main (#37336179188)
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