Moritz Schauer

Hi @touste You could define ``` Kalman.llikelihood(yres, S) = Kalman.logpdf(Kalman.Gaussian(zero(yres), (S+S')/2), yres) ``` to force symmetry. But do you need actually the likelihood? If not, I could make a version...

master now follows your suggestion

This could be a bug in my code or a bug in Julia. So what version of Julia are you running? Can you report the output of the function ```...

I think this is a Julia bug. Perhaps you can try the recent Julia 1.5.3? You can download it at https://julialang.org/

Here you are: ``` import Base: eltype, iteratorsize, iteratoreltype, start, next, done, SizeUnknown, length, size immutable Peek{I} it::I end peek(itr) = Peek(itr) eltype{I}(::Type{Peek{I}}) = eltype(I) iteratorsize{I}(::Type{Peek{I}}) = iteratorsize(I) iteratoreltype{I}(::Type{Peek{I}}) =...

As topping, for iterators which have the next element in their state already (so nothing is really to be done), define e.g. ``` start{T}(r::Peek{UnitRange{T}}) = start(r.it) next{T}(r::Peek{UnitRange{T}}, i) = next(r.it,...

I think `!isnull && done == last` so you could get rid of `last`

PS: last && return val, (s, val, false) and return done(f.it, s) && !last were not typos, your changes to done and next break the iterator: ``` julia> for i...

@dfdx It kind of "worked" because you provided the right "length" and julia's comprehension does not call "done" in that particular case

That's what I meant. This iterator is actually quite neat.