Tamas K. Papp

@cscherrer: thanks for asking, and the original answer. I understand that this is sometimes done, but at the moment I am not convinced that the optimization is worth the extra...

I played around with random tests and the new implementation is sometimes better (**much** better) and sometimes worse than `sum(log, ...)`. It would be great to understand why and see...

@mcabbott: thanks for the correction, indeed I didn't use the RNG. I think you are right. You can make a pathological example with eg ```julia z = [1 + (eps(T)...

@cscherrer: Regarding stepping back and agreement: I always think in terms of costs and benefits (code complexity and maintainability vs how useful the code is), and personally I would just...

AFAIK the only way to do this for loess is with bootstrapping.

AFAIK `FIELDS` is for documenting fields of a composite type (`struct`, `mutable struct`), not function arguments. Typically one does something like ```julia """ $(SIGNATURES) # Arguments - `startingAmount`: how much...

I am not a core dev, but I am not sure that the issue linked above is a sufficiently strong reason to even consider moving something into Base. Technically, there...

Yes and no. In a lot of cases resolving that would help me. Eg ```julia """ $(TYPEDSIGNATURES) """ foo(x, ::Val{N}) where N= nothing ``` does precisely what I want, but...

Revisiting this issue, I am wondering about the utility of this abbreviation, as subtypes of a type is an essentially dynamic value that may change when new code is evaluated...

I am kind of stuck with this one, eg ```julia julia> using DocStringExtensions julia> function f((x, y)) x + y end f (generic function with 1 method) julia> ms =...