Kristoffer Carlsson

icon indicating a website link https://kristofferc.github.io/ icon indicating an email [email protected]

icon company @JuliaComputing icon location Sweden icon location DM me on JuliaLang Slack for urgent things.

Results 1581 comments of Kristoffer Carlsson

Add new public function `Pkg.satisfies_compat(::VersionNumber, ::String)`, and add (to docs) some examples of checking `[compat]` entries

But if we go with that there is no need for the `satisfies_compat` method at all?

Add new public function `Pkg.satisfies_compat(::VersionNumber, ::String)`, and add (to docs) some examples of checking `[compat]` entries

A VersionSpec can contain a collection of ranges.

Add new public function `Pkg.satisfies_compat(::VersionNumber, ::String)`, and add (to docs) some examples of checking `[compat]` entries

A VersionSpec is a single object (that can internally have many ranges in it) , that you can test `in` with a VersionNumber with.

Add new public function `Pkg.satisfies_compat(::VersionNumber, ::String)`, and add (to docs) some examples of checking `[compat]` entries

Ah sorry, I should have been more accurate and said `VersionSpec`.

collecting a tree iterator is horribly type unstable, with lots of run time dispatch, even when the element type is provided explicitly

I think this is because it wants you to define a `childstatetype`: ``` """ childstatetype(::Type{T}) childstatetype(n) Indicates the type of the iteration state of the tree node `n` or its...

parse! with an IOBuffer errors on Julia 1.11-rc1

> I'm wondering, though, whether the root cause here is some breaking change in Julia itself? Nah, it grabs the internal `.data` field and assumes it is a `Vector{UInt8}`: https://github.com/JuliaLang/JuliaSyntax.jl/blob/3bf262bba32e833ed6d0d59a455a62faee97b408/src/parse_stream.jl#L318

Diagnostics are too chatty

You can chose what to propagate, or? Analogous to https://github.com/JuliaLang/JuliaSyntax.jl/pull/344

Regression in 1.0, asking for Hessian of a complex matrix function results in infinite loop

I'd suggest reading though https://github.com/JuliaDiff/ForwardDiff.jl/issues/480 and adding your comment there instead of rehashing the discussion here again.

Regression in 1.0, asking for Hessian of a complex matrix function results in infinite loop

> Can this regression be fixed after all? The current behavior is more correct. Just as an indication, you can see the number of issues of wrong gradients that were...

Regression in 1.0, asking for Hessian of a complex matrix function results in infinite loop

> The old behaviour was definitely correct, I feel you are missing the forest because of the tree. Just because there is one case that worked with the old behavior...