I think it's related to some old Matlab convention that Julia inherited, where there is an implicit infinity of singleton dimensions:
rand(5,5)[2,2,1,1] # this is valid rand(5,5)[1,:] # Returns a 1x5 2D array rand(5,5)[:,1] # returns a 5-element 1D array It's kinda consistent in its own way, but it'll change in in Julia 0.5. There's infinity-discussion about it on github <https://github.com/JuliaLang/julia/issues/13157>. On Tuesday, February 2, 2016 at 7:03:55 PM UTC-5, Davide Lasagna wrote: > > Hi, > > Just out of curiosity. What is the fundamental reason why > > size(rand(5, 5), 3) > > is equal to 1 and does not raise an error? > > Davide > > >
