Can't precompile on Julia v1.11.0-beta1

[frankebel]

If I try to load the package with include SparseIR on Julia v1.11.0-beta1, I get an error that the package does not precompile. Looking at the stacktrace, it comes from:

Stacktrace:
  [4] cosh(x::MultiFloats.MultiFloat{Float64, 2})
    @ SparseIR src/SparseIR.jl:21
 [12] segments_x(hints::SparseIR.SVEHintsLogistic{Float64}, ::Type{MultiFloats.MultiFloat{Float64, 2}})
    @ SparseIR src/kernel.jl:204

Problem seems to this block.

[shinaoka]

Thank you for the report. I have quickly reproduced the error on my Macbook Air M1.

PkgPrecompileError: The following 1 direct dependency failed to precompile:

SparseIR

Failed to precompile SparseIR [4fe2279e-80f0-4adb-8463-ee114ff56b7d] to "/Users/hiroshi/.julia/compiled/v1.11/SparseIR/jl_DTRTGY".
ERROR: LoadError: MethodError: no method matching _precision_with_base_2(::Type{MultiFloats.MultiFloat{Float64, 2}})
The function `_precision_with_base_2` exists, but no method is defined for this combination of argument types.

Closest candidates are:
  _precision_with_base_2(::BigFloat)
   @ Base mpfr.jl:957
  _precision_with_base_2(::Type{BigFloat})
   @ Base mpfr.jl:962
  _precision_with_base_2(::Type{Float64})
   @ Base float.jl:859
  ...

Stacktrace:
  [1] _precision(x::Type, base::Int64)
    @ Base ./float.jl:862
  [2] precision(::Type{MultiFloats.MultiFloat{Float64, 2}}; base::Int64)
    @ Base ./float.jl:865
  [3] precision(::Type{MultiFloats.MultiFloat{Float64, 2}})
    @ Base ./float.jl:865
  [4] cosh(x::MultiFloats.MultiFloat{Float64, 2})
    @ SparseIR ~/.julia/packages/SparseIR/RfK9H/src/SparseIR.jl:21
  [5] _broadcast_getindex_evalf
    @ ./broadcast.jl:673 [inlined]
  [6] _broadcast_getindex
    @ ./broadcast.jl:646 [inlined]
  [7] _getindex
    @ ./broadcast.jl:670 [inlined]
  [8] _broadcast_getindex
    @ ./broadcast.jl:645 [inlined]
  [9] getindex
    @ ./broadcast.jl:605 [inlined]
 [10] copy
    @ ./broadcast.jl:906 [inlined]
 [11] materialize
    @ ./broadcast.jl:867 [inlined]
 [12] segments_x(hints::SparseIR.SVEHintsLogistic{Float64}, ::Type{MultiFloats.MultiFloat{Float64, 2}})
    @ SparseIR ~/.julia/packages/SparseIR/RfK9H/src/kernel.jl:204
 [13] segments_x(hints::SparseIR.SVEHintsReduced{SparseIR.SVEHintsLogistic{Float64}}, ::Type{MultiFloats.MultiFloat{Float64, 2}})
    @ SparseIR ~/.julia/packages/SparseIR/RfK9H/src/kernel.jl:395
 [14] SparseIR.SamplingSVE(kernel::SparseIR.ReducedKernel{SparseIR.LogisticKernel}, ε::Float64, ::Type{MultiFloats.MultiFloat{Float64, 2}}; n_gauss::Nothing)
    @ SparseIR ~/.julia/packages/SparseIR/RfK9H/src/sve.jl:42
 [15] SparseIR.CentrosymmSVE(kernel::SparseIR.LogisticKernel, ε::Float64, ::Type{MultiFloats.MultiFloat{Float64, 2}}; InnerSVE::Type{SparseIR.SamplingSVE}, n_gauss::Nothing)
    @ SparseIR ~/.julia/packages/SparseIR/RfK9H/src/sve.jl:85
 [16] SparseIR.SVEResult(kernel::SparseIR.LogisticKernel; Twork::Nothing, cutoff::Nothing, ε::Nothing, lmax::Int64, n_gauss::Nothing, svd_strat::Symbol, SVEstrat::Type{SparseIR.CentrosymmSVE})
    @ SparseIR ~/.julia/packages/SparseIR/RfK9H/src/sve.jl:158
 [17] SparseIR.FiniteTempBasis(statistics::SparseIR.Fermionic, β::Float64, ωmax::Float64, ε::Nothing)
    @ SparseIR ~/.julia/packages/SparseIR/RfK9H/src/basis.jl:67
 [18] macro expansion
    @ ~/.julia/packages/SparseIR/RfK9H/src/SparseIR.jl:50 [inlined]
 [19] macro expansion
    @ ~/.julia/packages/PrecompileTools/L8A3n/src/workloads.jl:78 [inlined]
 [20] top-level scope
    @ ~/.julia/packages/SparseIR/RfK9H/src/SparseIR.jl:49
 [21] include
    @ ./Base.jl:558 [inlined]
 [22] include_package_for_output(pkg::Base.PkgId, input::String, depot_path::Vector{String}, dl_load_path::Vector{String}, load_path::Vector{String}, concrete_deps::Vector{Pair{Base.PkgId, UInt128}}, source::Nothing)
    @ Base ./loading.jl:2721
 [23] top-level scope
    @ stdin:4
in expression starting at /Users/hiroshi/.julia/packages/SparseIR/RfK9H/src/SparseIR.jl:1
in expression starting at stdin:

[shinaoka]

@SamuelBadr @mwallerb

The function _precision_with_base_2 in base/float.jl was introduced recently.

Can we work around this by overwriting the function precision here?

https://github.com/SpM-lab/SparseIR.jl/blob/main/src/SparseIR.jl#L21

[SamuelBadr]

Thanks both!

The new version 1.0.18 fixes this.