So CI reveals out this native implementation can report (potentially too many) false positives.

julia> @profile_call sin(1)
═════ 1 possible error found ═════
┌ @ special/trig.jl:53 Base.Math.sin(Base.Math.float(x::Int64)::Float64)
│┌ @ special/trig.jl:39 Base.Math.sin_domain_error(x::Float64)
││┌ @ special/trig.jl:28 unreachable
│││ will throw: Base.Math.throw(Base.Math.DomainError(x::Float64, "sin(x) is only defined for finite x.")::Any)
││└──────────────────────
Float64

These lines of code suppress throw calls that may not happen by analyzing the current inference frame, but what we need to address the case above is (more complicated) inter-frame analysis.

EDIT: simple false positives are fixed by simple inter-frame analysis

Codecov Report

Merging #17 into master will increase coverage by 2.43%.
The diff coverage is 96.00%.

@@            Coverage Diff             @@
##           master      #17      +/-   ##
==========================================
+ Coverage   65.77%   68.21%   +2.43%     
==========================================
  Files          10       10              
  Lines         561      604      +43     
==========================================
+ Hits          369      412      +43     
  Misses        192      192              

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

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 73be65f...d47959c. Read the comment docs.