Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (444331d) 90.76% compared to head (424e062) 90.75%.

@@            Coverage Diff             @@
##           master     #248      +/-   ##
==========================================
- Coverage   90.76%   90.75%   -0.02%     
==========================================
  Files          23       23              
  Lines        2122     2119       -3     
==========================================
- Hits         1926     1923       -3     
  Misses        196      196              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

I hope I'm not abusing your goodwill, but could you explain why you changed LinearAlgebra.dot(lhs::A, rhs::B) to @eval LinearAlgebra.dot(lhs::$A, rhs::$B)? As far as I understand both alternatives do the same thing.

It would be necessary if A were a symbol, like :Hermitian, but A is just the type directly, Hermitian.

It's really just a personal preference thing. I like that @eval is a signal that we're generating a bunch of new methods. If it's just lhs::A, then (I think) the code is a little more likely to be glossed over.

Ok, thanks for the answer. I was just curious because in LinearAlgebra they defined A as :Hermitian, I was afraid there was a deep reason to prefer one over the other.