in particular, this code seems really weird:
|
for (M, f, arity) in DiffRules.diffrules(filter_modules = nothing) |
|
if (M, f) in ((:Base, :^), (:NaNMath, :pow), (:Base, :/), (:Base, :+), (:Base, :-), (:Base, :sin), (:Base, :cos)) |
|
continue # Skip methods which we define elsewhere. |
|
elseif !(isdefined(@__MODULE__, M) && isdefined(getfield(@__MODULE__, M), f)) |
|
continue # Skip rules for methods not defined in the current scope |
|
end |
|
if arity == 1 |
|
eval(unary_dual_definition(M, f)) |
|
elseif arity == 2 |
|
eval(binary_dual_definition(M, f)) |
|
else |
|
# error("ForwardDiff currently only knows how to autogenerate Dual definitions for unary and binary functions.") |
|
# However, the presence of N-ary rules need not cause any problems here, they can simply be ignored. |
|
end |
|
end |
in particular because NaNMath is loaded beforehand:
The same thing can be said about SpecialFunctions.
I noticed that DiffRules also depends on NaNMath and SpecialFunctions, but does not load those packages. ForwardDiff does.
