Thanks Moritz for this!

Interpolation is a huge subject, I understand your choices (Left, Right, and Linear), but it might be better to use the Interpolations.jl package for this kind of work? Alternatively, you could make sure it works in Interpolations.jl, show an example, but exclude the code from your package to keep it light...

Interpolations in trajectories is tightly coupled with smoothing as well: if you want to interpolate you might want to smooth as well. So you can see how this becomes much larger (with splines or a Kalman filter). I think you might run the risk of including everything but the kitchen sink... So again, it might be better to include an example with Dierckx and your Kalman filter, but exclude the actual packages from your dependencies.

Just some thoughts.

The tag name "v0.1" is not of the appropriate SemVer form (vX.Y.Z).
cc: @mschauer

In Julia, destructuring defaults to iteration.

Unfortunately that means

t, x = X

and

collect(X) = [t1 => x1, t2=>x2]

or collect(X) = [x1, x2, ...] cannot coexist. Current workaround is to demand explicit

t, x = Pair(X)
collect(pairs(X))
collect(values(X))

@JuliaRegistrator register

I am planning to use Kalman.jl in a project which uses Tables.jl v1.0, and the released version of Trajectories.jl prevents that from being installed.

(Works with #master installed)

The @unroll1 macro does not introduce a new scope. Example:

function only(x)
    Trajectories.@unroll1 for z in x
         if !$first
               throw(ArgumentError("too many elements"))
           end
    end
    return z
end