Julia 1.9 error: expected Static.StaticInt{1}, got a value of type Static.StaticInt{0}

[droodman]

I hit an error moving from 1.8.5 to 1.9.0. In the MWE below, I create a 1-vector of vectors. I then run a @turbo loop referring to the first (only) entry in this 1-vector, which I index to using the expression 1+false. It crashes in 1.9.0 but not 1.8.5. Is this a LoopVectorization issue, or is my on-the-fly mixing of integer and Boolean bad enough Julia style that I deserve whatever I get?

There is an easy work-around, which is to wrap the false in Int().

julia> versioninfo()
Julia Version 1.9.0
Commit 8e63055292 (2023-05-07 11:25 UTC)
Platform Info:
  OS: Windows (x86_64-w64-mingw32)
  CPU: 16 × Intel(R) Core(TM) i7-10875H CPU @ 2.30GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-14.0.6 (ORCJIT, skylake)
  Threads: 1 on 16 virtual cores
Environment:
  JULIA_IMAGE_THREADS = 1

julia> using LoopVectorization

julia> u = [rand(2)]
1-element Vector{Vector{Float64}}:
 [0.26563842305190977, 0.30181657779394844]

julia> @turbo for r ∈ indices(u[1+false],1)
         u[1][r] = 1
       end
ERROR: TypeError: in typeassert, expected Static.StaticInt{1}, got a value of type Static.StaticInt{0}
Stacktrace:
 [1] convert
   @ .\number.jl:7 [inlined]
 [2] _promote
   @ .\promotion.jl:358 [inlined]
 [3] promote
   @ .\promotion.jl:381 [inlined]
 [4] +(x::Static.StaticInt{1}, y::Bool)
   @ Base .\promotion.jl:410
 [5] top-level scope
   @ .\REPL[10]:1

julia> @turbo for r ∈ indices(u[1+Int(false)],1)
         u[1][r] = 1
       end

julia>

[droodman]

Oh the problem is gone after, up in the package manager. Maybe updating to the latest Static solves the problem? Static v0.8.6 ⇒ v0.8.7

[chriselrod]

Oh the problem is gone after, up in the package manager. Maybe updating to the latest Static solves the problem? Static v0.8.6 ⇒ v0.8.7

Yes, it should. Static 0.8.7 fixes broken type promotion/conversion in Static, that started causing problems in 1.9 as Julia 1.9 started adding asserts.

[kongdd]

Faced with the same issue: brenhinkeller/NaNStatistics.jl#35

[chriselrod]

Is this still a problem with the latest version of Static, or can I close this?

[droodman]

You can close it.