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

stillwater-sc / universal / 38960461117
89%
master: 84%

Build:
Build:
LAST BUILD BRANCH: main
DEFAULT BRANCH: master
Ran 16 Oct 2026 08:55PM UTC
Jobs 1
Files 723
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

16 Oct 2026 08:11PM UTC coverage: 86.745% (+0.01%) from 86.732%
38960461117

push

github

web-flow
fix(poxel): give tile_interval's limit corners their real limit (#1704)

tile_interval's combine() did not know which side of the interval an
end bounds, so an open zero had no direction. It fell back to the
widest bound or a closed 0 at every limit corner:

  [1,2] / (0,1]    gave (-inf, inf)  now [1, inf)
  (0,1] * (0,1]    gave [0, 1]       now (0, 1]
  (0,1] * [1,inf)  gave [0, inf)     now (0, inf)
  [1] / [1,inf)    gave [0, 1]       now (0, 1]

The old answers were sound but too wide. The corner logic now matches
ubound's (#1700): each end carries its side (+1 lower, -1 upper), an
open end is reached only through a closed-zero partner, x / 0+- takes
the sign sgn(x) * side, an open zero against an open zero or an
infinity spreads with sign xside * yside, and finite / inf is an open 0.

The enclosure tests could not see this, because (-inf, inf) encloses
everything. tile_interval.cpp adds a tightness check against an exact
rational oracle over every pair of finite tiles and over hulls, plus
explicit limit cases at infinite ends. The unfixed header fails the
check 759 to 2552 times per tile type, and mutating the strictness rule,
the divisor side or the open-zero product is caught. The test now uses
the standard MANUAL_TESTING / REGRESSION_LEVEL skeleton.

ucalc_regression held two expectations that relied on the loose answers:
- areal8 now decides the sign of the stable root r1 of 3x^2 + 100x + 2;
  its box is (-0.562, 0), where it used to be (-0.562, 0].
- rootbox 1/x on [-1, 1] now proves there is no root; the tiles beside
  the pole used to be reported as undecided.

Resolves #1702

BEHAVIOUR-CHANGE: tile_interval (areal/poxel) * and / at open-zero or infinite ends return tighter intervals; ucalc rootbox and precision-sweep results change accordingly

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

30 of 31 new or added lines in 1 file covered. (96.77%)

49313 of 56848 relevant lines covered (86.75%)

12121061.76 hits per line

Uncovered Changes

Lines Coverage ∆ File
1
98.58
-0.46% include/sw/universal/utility/tile_interval.hpp
Jobs
ID Job ID Ran Files Coverage
1 38960461117.1 16 Oct 2026 08:55PM UTC 723
86.75
GitHub Action Run
Source Files on build 38960461117
  • Tree
  • List 723
  • Changed 2
  • Source Changed 1
  • Coverage Changed 2
Coverage ∆ File Lines Relevant Covered Missed Hits/Line
  • Back to Repo
  • Github Actions Build #38960461117
  • a84f1a89 on github
  • Prev Build on main (#38930407030)
  • Next Build on main (#38983239869)
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