Cannot specify type of list comprehension

[LilithHafner]

Zygote can differentiate list comprehensions with automatically generated types ([i+1 for i in 1:10]), but not for explicitly specified types (Int[i+1 for i in 1:10] or Float64[i+1 for i in 1:10]).

MWE

julia> gradient(1.0) do x
           sum([x for _ in 1:10])
       end
(10.0,)

julia> gradient(1.0) do x
           sum(Float64[x for _ in 1:10])
       end
ERROR: Mutating arrays is not supported -- called setindex!(Vector{Float64}, ...)
This error occurs when you ask Zygote to differentiate operations that change
the elements of arrays in place (e.g. setting values with x .= ...)

Possible fixes:
- avoid mutating operations (preferred)
- or read the documentation and solutions for this error
  https://fluxml.ai/Zygote.jl/latest/limitations

Stacktrace:
 [1] error(s::String)
   @ Base ./error.jl:35
 [2] _throw_mutation_error(f::Function, args::Vector{Float64})
   @ Zygote ~/.julia/packages/Zygote/SmJK6/src/lib/array.jl:86
 [3] (::Zygote.var"#389#390"{Vector{Float64}})(#unused#::Nothing)
   @ Zygote ~/.julia/packages/Zygote/SmJK6/src/lib/array.jl:98
 [4] (::Zygote.var"#2509#back#391"{Zygote.var"#389#390"{Vector{Float64}}})(Δ::Nothing)
   @ Zygote ~/.julia/packages/ZygoteRules/AIbCs/src/adjoint.jl:67
 [5] Pullback
   @ ./REPL[49]:2 [inlined]
 [6] (::typeof(∂(#47)))(Δ::Float64)
   @ Zygote ~/.julia/packages/Zygote/SmJK6/src/compiler/interface2.jl:0
 [7] (::Zygote.var"#60#61"{typeof(∂(#47))})(Δ::Float64)
   @ Zygote ~/.julia/packages/Zygote/SmJK6/src/compiler/interface.jl:45
 [8] gradient(f::Function, args::Float64)
   @ Zygote ~/.julia/packages/Zygote/SmJK6/src/compiler/interface.jl:97
 [9] top-level scope
   @ REPL[49]:1

julia> VERSION
v"1.9.0-alpha1"

(jl_NHms0y) pkg> status
Status `/private/var/folders/hc/fn82kz1j5vl8w7lwd4l079y80000gn/T/jl_NHms0y/Project.toml`
  [e88e6eb3] Zygote v0.6.51

[ToucheSir]

julia> Meta.@lower Int[i+1 for i in 1:10]
:($(Expr(:thunk, CodeInfo(
    @ none within `top-level scope`
1 ─ %1  = 1:10
│   %2  = Base.IteratorSize(%1)
│   %3  = %2 isa Base.SizeUnknown
│   %4  = Base._array_for(Int, %1, %2)
│   %5  = Base.LinearIndices(%4)
│         @_2 = Base.first(%5)
│         #s1 = Base.iterate(%1)
│   %8  = #s1 === nothing
│   %9  = Base.not_int(%8)
└──       goto #7 if not %9
2 ┄ %11 = #s1
│         i = Core.getfield(%11, 1)
│   %13 = Core.getfield(%11, 2)
│   %14 = i + 1
│         nothing
└──       goto #4 if not %3
3 ─       Base.push!(%4, %14)
└──       goto #5
4 ─       Base.setindex!(%4, %14, @_2)
5 ┄       nothing
│         @_2 = Base.add_int(@_2, 1)
│         #s1 = Base.iterate(%1, %13)
│   %23 = #s1 === nothing
│   %24 = Base.not_int(%23)
└──       goto #7 if not %24
6 ─       goto #2
7 ┄       return %4
))))

Not much we can do when the compiler unconditionally lowers to code using mutation, unfortunately.

[LilithHafner]

#75 would be a great fix :)

[ToucheSir]

There's also an infinitesimally small chance it'll happen as things stand with Zygote development. The only reason I haven't added the wontfix label here is in the off change the lowering for that syntax might change in some future Julia version. For now, we can use GitHub's suggested verbiage and say this is "not planned".

[LilithHafner]

Is the plan to not add support for mutation? (duplicate of #75 (comment))

[ToucheSir]

There are no plans until someone comes up with a proposal/PR that overcomes everything mentioned in #1228 (comment). Thus far this has not happened, and so nothing is planned.