[Rafael-SOWNet]

1/((x^2)^tanh(1)) at 0 was NaN, the claim that the limit does not exist. The limit is +oo from both sides.

IsNeverNegative only counted a base as never negative when it was an absolute value, or a real power, product or quotient of never-negative parts. x^2 was none of those. So a vanishing power of it was left unevaluated from each side when its exponent was a constant other than a number, or tended to a negative number. The two-sided limit was then NaN because neither side had an answer. 2.6.0 gives the same results.

An even integer power is now never negative when TreeAnalyzer.IsRealValued says its base is real for real x. IsNeverNegative now takes the variable so it can ask that. The base has to be real, because sqrt(x)^2 is x. 1/(((i*x)^2)^tanh(1)) has base -x^2, and it is still not answered. A test pins that.

Input	2.6.0 and master	Now
1/((x^2)^tanh(1)) at 0	NaN; each side unevaluated	+oo
(sinh(x)^2)^(sin(x) - 1) at 0	NaN; each side unevaluated	+oo
sqrt(x^2)^tan(pi - 1) at 0	NaN; each side unevaluated	+oo

Measured

• limitcheck at DEPTH=3 on 11 seeds (14, 16, 19, 23, 26, 28, 29, 31, 33, 34, 36), master against this branch: three NaNs are answered, all +oo. Two are at seed 36, 1/((x^2)^tanh(1)) and (sinh(x)^2)^(sin(x) - 1). One is at seed 29, sqrt(x^2)^tan(pi - 1). Nothing else changed.
• The new tests fail on master: all 6 rows of the theory fail, and the guard on the complex base passes.
• Suite: 18963 passed, 0 failed. AOT passed. BoundCheck, RootCheck, CasBench, PropCheck, SimpSweep, RuleCheck, CanonCheck and Confluence are clean.

Recorded in BREAKING-CHANGES.md.

🤖 Generated with Claude Code

[codecov]

Codecov Report

❌ Patch coverage is 81.81818% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 86.78%. Comparing base (984671d) to head (a98cc6f).
⚠️ Report is 2 commits behind head on master.

Files with missing lines	Patch %	Lines
...ns/Continuous/Limits/Solvers/Solvers.Definition.cs	0.00%	0 Missing and 1 partial ⚠️
...ath/Functions/Continuous/Limits/Transformations.cs	88.88%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #2368   +/-   ##
=======================================
  Coverage   86.78%   86.78%           
=======================================
  Files         312      312           
  Lines       68861    68865    +4     
  Branches    20066    20067    +1     
=======================================
+ Hits        59761    59765    +4     
+ Misses       4911     4910    -1     
- Partials     4189     4190    +1     

Flag	Coverage Δ
cpp	37.77% <ø> (ø)
csharp	87.32% <81.81%> (+<0.01%)	⬆️
fsharp	32.97% <ø> (ø)
interactive	40.72% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
...nctions/Continuous/Limits/Solvers/Limit.Classes.cs	72.30% <100.00%> (-1.08%)	⬇️
...ns/Continuous/Limits/Solvers/Solvers.Definition.cs	89.40% <0.00%> (-0.47%)	⬇️
...ath/Functions/Continuous/Limits/Transformations.cs	88.37% <88.88%> (+0.07%)	⬆️

... and 9 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Rafael-SOWNet]

Merged without the two macOS jobs. They never ran: GitHub had no macOS arm64 runner capacity, and each of three re-runs was cancelled with "The job was not acquired by Runner of type hosted even after multiple attempts". No macOS job has started on any branch since 14:31 UTC. In their place:

• I ran every row of AnEvenPowerVanishesFromAboveTest on a 512 KB thread, the tightest stack the macOS leg gives. All 21 limits came out as the tests expect.
• Every other check is green, on Linux and Windows.
• The local chain, suite and harnesses all passed.