In 0.6 we deprecated isinteger(a::AbstractArray) to all(isinteger, a). However we still have such a method for isreal.
If any and all had inbounds annotations, then we would be able to optimize these to constants.
I believe that was discussed somewhere and IIRC @stevengj was against that change for mathematical reasons.
Oh, ok. Then perhaps this release is converging after all :)
Yes, I still feel the same way.
Don't the same linear algebra considerations apply to both isreal and isinteger? I.e. shouldn't we deprecate both or none of them?
@nalimilan, no. You wouldn't say that [1,2,3] "is an integer". You would say that it "is real". There is a well defined, standard, sense in which operations like real and conj apply to vectors as a whole, and not just to their contents.
Most helpful comment
@nalimilan, no. You wouldn't say that
[1,2,3]"is an integer". You would say that it "is real". There is a well defined, standard, sense in which operations likerealandconjapply to vectors as a whole, and not just to their contents.