Method ordering error

[kmsquire]

Noted here:

julia> Pkg.test("GZip")
INFO: Testing GZip
WARNING: New definition
    write(GZip.GZipStream, Array{#T<:Any, N<:Any}) at /Users/kmsquire/.julia/v0.5/GZip/src/GZip.jl:456
is ambiguous with:
    write(Base.IO, Array{UInt8, N<:Any}) at io.jl:154.
To fix, define
    write(GZip.GZipStream, Array{UInt8, N<:Any})
before the new definition.

This seems like an bug. GZipStream is a concrete subclass of IO, so shouldn't the definition involving it take precedence?

julia> versioninfo()
Julia Version 0.5.0-dev+2440
Commit 2bb94d6 (2016-02-01 02:22 UTC)
Platform Info:
  System: Darwin (x86_64-apple-darwin13.4.0)
  CPU: Intel(R) Core(TM)2 Extreme CPU X7900  @ 2.80GHz
  WORD_SIZE: 64
  BLAS: libopenblas (USE64BITINT DYNAMIC_ARCH NO_AFFINITY Core2)
  LAPACK: libopenblas64_
  LIBM: libopenlibm
  LLVM: libLLVM-3.7.1

[yuyichao]

Why is this a bug? The first method only works with GZipStream and the second method only works with Array{UInt8} so they are indeed ambiguous?

[yuyichao]

In another word, write(::GZip.GZipStream, ::Vector{Int}) only matches the first one and write(::IOBuffer, ::Vector{UInt8}) only matches the second one so none of them are more specific than the other so they are ambiguous.

I remember there's some discussion about sorting the method using the first argument first but I don't think we have that yet.

[kmsquire]

Hmm.. Okay.

I see that this was caused by #14766, so GZip.jl will have to be updated to deal with that then.

[carnaval]

I remember there's some discussion about sorting the method using the first argument first but I don't think we have that yet.

as an aside, please no. the fact that our multiple dispatch is symmetric is to me what makes it elegant

[kmsquire]

No worries--I think it was just my misunderstanding. Cheers!