Sebastian Pfitzner

This very much looks like an UInt16 overflow, since ``` julia> fs[2](1.0) == fs[65538](1.0) true julia> fs[3](1.0) == fs[65539](1.0) true julia> fs[4](1.0) == fs[65540](1.0) true ``` 65532-65537 aren't callable: ```...

Jeff suggested running a debug build to check whether this code trips any assertions, and indeed, it does: ``` julia> calc(fs) 10000 20000 30000 40000 50000 60000 julia-debug: /home/pfitzseb/Documents/Git/julia/src/ircode.c:344: jl_encode_value_:...

https://github.com/JuliaLang/julia/pull/45173 fixes this for (at least) n = 2e5.

I feel like I'm missing some context here :) Can't you just insert LNNs to where the expression is defined or something?

Ok, I was just going by the example in the readme. Makes sense that that doesn't reflect real usecases though. It should be fairly easy to support the `Ctrl-Q` case...

Should be fixed by re-generating the docs after https://github.com/JuliaRegistries/General/pull/73947 is merged.

Yeah, I suppose we can do that and look at the iterator interface instead of the AbstractArray interface, which does in fact requires `size` to be implemented: https://docs.julialang.org/en/v1/manual/interfaces/#man-interface-array.

Yeah, that's a known limitation atm.

You can always disable the extension for the workspace specifically. That said, the extension should only activate automatically if there's a `Project.toml` in your workspace, you open a Julia file,...

We could just be a bit more deliberate about when we start the LS. Extension activation doesn't *have to* start the LS, but does currently in the Julia extension.