Neven Sajko

AbstractTrees.jl support

1

https://juliahub.com/ui/Packages/General/AbstractTrees The only two functions that are mandatory to overload are `children` and `nodevalue`.

How to handle functions which run slowly the first few thousand times and fast on subsequent runs?

3

In a fresh REPL session: ```julia foo(::Type{ foo(::Type{ T = Vector{Int}; julia> g() = foo(T) g (generic function with 1 method) julia> using Chairmarks julia> @b g 333.609 ns julia>...

fix a huge part of invalidation on defining a new `Integer` subtype

5

I have no idea why this helps, it really seems like Julia should be able to infer the type of `n` as `Int` on its own. Perhaps something to do...

`Integer`s in the type domain

12

Introduce `TypeDomainInteger`, a subtype of `Integer` in the type domain. That is, each value is of singleton type. These type domain integers are then applied to many places in `Base`...

slightly better type inference for `reverse(::NTuple)`

17

Just adds an `::NTuple` type assertion to the result. Before the return type of `reverse(::NTuple)` inferred as `Tuple`, now it's more precise, inferring as `NTuple`. This is a partial reland...

hashing is part unimplemented, part incorrect

5

Test: ```julia using Test, Infinities @testset "hash versus floating-point infinity" begin @testset "one argument" begin @test hash(Inf) === hash(∞) end @testset "two arguments" begin for h ∈ rand(UInt, 256) @test...

The README has links to package docs and tutorial, but the links lead to 404 pages.

broken links to package documentation and tutorial

1

The README has links to package docs and tutorial, but the links lead to 404 pages.

This is the method: https://github.com/JuliaLang/julia/blob/877de9839809e769d4f9707a61df3400d087d8d3/base/rational.jl#L101-L101 The math behind the transformation incorrectly assumes `y` is finite and nonzero. xref #53435, which concerns overflow caused by the same oneliner. Test: ```julia using...

implement `sind` in terms of `sinpi` instead of in terms of `sin`, same for `cosd`

1

I guess this would result in better accuracy and/or performance. Current implementations: https://github.com/JuliaLang/julia/blob/fe864374d50b121aeb3ce5473c206338b21df5e5/base/special/trig.jl#L1209-L1264