JuliaLang / JuliaLang/Distributed.jl
Type inference fails for `@distributed for` but not for `mapreduce`
Open
Nobody has claimed this yet.
- Dominant language
- Julia
- Stars
- 55
- Forks
- 19
- PR merge metrics
- No merged PRs in 30d
Description
In some cases, type inference works for mapreduce but fails for @distributed for.
function foo()
A = Array{Float64}(undef, 10)
result = @distributed (+) for _ in 1:100
rand!(A)
sum(A)
end
end
function bar()
A = Array{Float64}(undef, 10)
result = mapreduce(+, 1:100) do _
rand!(A)
sum(A)
end
end
julia> @code_warntype foo()
Variables
#self#::Core.Const(foo)
JuliaLang/julia#272::var"#272#273"{Vector{Float64}}
result::Any
A::Vector{Float64}
Body::Any
1 ─ %1 = Core.apply_type(Main.Array, Main.Float64)::Core.Const(Array{Float64, N} where N)
│ (A = (%1)(Main.undef, 10))
│ %3 = Main.:(var"#272#273")::Core.Const(var"#272#273")
│ %4 = Core.typeof(A)::Core.Const(Vector{Float64})
│ %5 = Core.apply_type(%3, %4)::Core.Const(var"#272#273"{Vector{Float64}})
│ (#272 = %new(%5, A))
│ %7 = JuliaLang/julia#272::var"#272#273"{Vector{Float64}}
│ %8 = (1:100)::Core.Const(1:100)
│ %9 = Distributed.preduce(Main.:+, %7, %8)::Any
│ (result = %9)
└── return %9
julia> @code_warntype bar()
Variables
#self#::Core.Const(bar)
JuliaLang/julia#274::var"#274#275"{Vector{Float64}}
result::Float64
A::Vector{Float64}
Body::Float64
1 ─ %1 = Core.apply_type(Main.Array, Main.Float64)::Core.Const(Array{Float64, N} where N)
│ (A = (%1)(Main.undef, 10))
│ %3 = Main.:(var"#274#275")::Core.Const(var"#274#275")
│ %4 = Core.typeof(A)::Core.Const(Vector{Float64})
│ %5 = Core.apply_type(%3, %4)::Core.Const(var"#274#275"{Vector{Float64}})
│ (#274 = %new(%5, A))
│ %7 = JuliaLang/julia#274::var"#274#275"{Vector{Float64}}
│ %8 = (1:100)::Core.Const(1:100)
│ %9 = Main.mapreduce(%7, Main.:+, %8)::Float64
│ (result = %9)
└── return %9
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Reproduce the issue with the foo and bar examples in the report, then compare their @code_warntype output. Start by tracing the @distributed for path through Distributed.preduce and contrast it with mapreduce. Done means the distributed example no longer produces an Any result when the equivalent mapreduce inference is Float64.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- julia
- Domain
- distributed-systems
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100