In Flux0.10.4 Warning: Zygote does not have CuArrays in its dependencies

Specific Domains Machine Learning
1

from a fresh Julia1.4, adding the new Flux 0.10.4, followed by adding Zygote,
results in a warning

Package Zygote does not have CuArrays in its dependencies
…you may need to report an issue with Zygote.

Is this something to worry about?

2

In fact there is a stacktrace on the mac (which does not use cuda)

│ ArgumentError: Package Zygote does not have CuArrays in its dependencies:
│ - If you have Zygote checked out for development and have
│   added CuArrays as a dependency but haven't updated your primary
│   environment's manifest file, try `Pkg.resolve()`.
│ - Otherwise you may need to report an issue with Zygote
│ Stacktrace:
│  [1] require(::Module, ::Symbol) at ./loading.jl:905
│  [2] top-level scope at /Users/myuser/.julia/packages/Zygote/wkc82/src/lib/broadcast.jl:183
│  [3] eval at ./boot.jl:331 [inlined]
│  [4] eval at /Users/myuser/.julia/packages/Zygote/wkc82/src/Zygote.jl:1 [inlined]
│  [5] (::Zygote.var"#1769#1773")() at /Users/myuser/.julia/packages/Requires/qy6zC/src/require.jl:85
│  [6] err(::Any, ::Module, ::String) at /Users/myuser/.julia/packages/Requires/qy6zC/src/require.jl:42
│  [7] (::Zygote.var"#1768#1772")() at /Users/myuser/.julia/packages/Requires/qy6zC/src/require.jl:84
│  [8] withpath(::Any, ::String) at /Users/myuser/.julia/packages/Requires/qy6zC/src/require.jl:32
│  [9] (::Zygote.var"#1767#1771")() at /Users/myuser/.julia/packages/Requires/qy6zC/src/require.jl:83
│  [10] #invokelatest#1 at ./essentials.jl:712 [inlined]
│  [11] invokelatest at ./essentials.jl:711 [inlined]
│  [12] foreach at ./abstractarray.jl:1919 [inlined]
│  [13] loadpkg(::Base.PkgId) at /Users/myuser/.julia/packages/Requires/qy6zC/src/require.jl:22
│  [14] #invokelatest#1 at ./essentials.jl:712 [inlined]
│  [15] invokelatest at ./essentials.jl:711 [inlined]
│  [16] _tryrequire_from_serialized(::Base.PkgId, ::UInt64, ::Nothing) at ./loading.jl:715
│  [17] _require_from_serialized(::String) at ./loading.jl:743
│  [18] _require(::Base.PkgId) at ./loading.jl:1039
│  [19] require(::Base.PkgId) at ./loading.jl:927
│  [20] require(::Module, ::Symbol) at ./loading.jl:922
│  [21] eval at ./boot.jl:331 [inlined]
│  [22] repleval(::Module, ::Expr) at /Users/myuser/.julia/packages/Atom/wlPiw/src/repl.jl:209
│  [23] (::Atom.var"#228#230"{Module})() at /Users/myuser/.julia/packages/Atom/wlPiw/src/repl.jl:236
│  [24] with_logstate(::Atom.var"#228#230"{Module}, ::Base.CoreLogging.LogState) at ./logging.jl:398
│  [25] with_logger at ./logging.jl:505 [inlined]
│  [26] evalrepl(::Module, ::String) at /Users/myuser/.julia/packages/Atom/wlPiw/src/repl.jl:227
│  [27] top-level scope at /Users/myuser/.julia/packages/Atom/wlPiw/src/repl.jl:271
│  [28] eval(::Module, ::Any) at ./boot.jl:331
│  [29] eval_user_input(::Any, ::REPL.REPLBackend) at /Users/julia/buildbot/worker/package_macos64/build/usr/share/julia/stdlib/v1.4/REPL/src/REPL.jl:86
│  [30] macro expansion at /Users/julia/buildbot/worker/package_macos64/build/usr/share/julia/stdlib/v1.4/REPL/src/REPL.jl:118 [inlined]
│  [31] (::REPL.var"#26#27"{REPL.REPLBackend})() at ./task.jl:358
└ @ Requires ~/.julia/packages/Requires/qy6zC/src/require.jl:44
3

The best place to go is indeed on Issues · FluxML/Zygote.jl · GitHub and open an issue. CuArrays is for GPU computing, if you don’t need it flux should work as intended for you. I think.