ponder-lab / Hybridize-Functions-Refactoring / #3055
85%
main: 86%

LAST BUILD BRANCH: document-test-selector-separators
DEFAULT BRANCH: main
Ran
Jobs 1
Files 42
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
coverage: 85.491%. Remained the same
#3055

Pull #945

github

khatchad 
Pin that a concrete extent loses to a symbolic twin at the same axis

`inferSpec` takes per-position consensus, not the more precise member:
`NumericDim(46).equals(SymbolicDim("?"))` is false, so a concrete extent
standing beside a placeholder at the same position is a disagreement,
and a disagreement wildcards the axis.

That is the contract an upstream precision improvement has to satisfy to
reach the specification surface. A fix that adds a concrete member
beside an existing placeholder changes nothing here, because the
placeholder still disagrees with it. Only a fix that eliminates the
placeholder moves the emitted axis. The distinction was readable in the
reduction but pinned nowhere, so it would have been rediscovered
downstream rather than from the suite.

Three tests rather than one, so the first asserts what it appears to.
Alone, a wildcarded trailing axis is equally consistent with the
reduction never keeping any extent: `testAgreeingConcreteDimsSurvive`
witnesses that the same extents survive once the twin is removed, and
`testDisagreeingConcreteDimsWildcardOnlyThatAxis` separates the twin
case from ordinary multi-context disagreement, where the disputed axes
wildcard and an agreed trailing extent survives.

Verified by ablation rather than by a green run: inverting the trailing
expectation fails with `expected:<D:Constant,46> but
was:<D:Symbolic,?>`, so the assertion discriminates.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01BHbypoDrGUGiBz9xZWt8nd
Pull Request #945: Pin that a concrete extent loses to a symbolic twin at the same axis

3612 of 4225 relevant lines covered (85.49%)

0.85 hits per line

Jobs
ID Job ID Ran Files Coverage
1 #3055.1 42
85.49
Source Files on build #3055