in 1.11, push! doesn't forward to append! as it should

[aplavin]

push! docstring:

push!(collection, items...) -> collection

Insert one or more items in collection. If collection is an ordered container, the items are inserted at the end (in the given order).

Examples
≡≡≡≡≡≡≡≡

julia> push!([1, 2, 3], 4, 5, 6)
6-element Vector{Int64}:
1
2
3
4
5
6

If collection is ordered, use append! to add all the elements of another collection to it. The result of the preceding example is equivalent to append!([1, 2, 3], [4, 5, 6]). For AbstractSet objects, union! can be used instead.

Naturally, one would expect that defining append!(::MyArray, xs) would enable push! as well. And it does – on 1.10 and below.
But 1.11 breaks this.

MWE:

struct MyArray <: AbstractVector{Int}
    data::Vector{Int}
end

Base.size(a::MyArray) = size(a.data)
Base.getindex(a::MyArray, i...) = getindex(a.data, i...)

function Base.append!(a::MyArray, b)
    append!(a.data, b)
    return a
end

a = MyArray([1, 2, 3])
@show a
push!(a, 4)
@show a

Results:

❯ julia +1.6 tmp.jl 
a = [1, 2, 3]
a = [1, 2, 3, 4]

❯ julia +1.10 tmp.jl
a = [1, 2, 3]
a = [1, 2, 3, 4]

❯ julia +1.11 tmp.jl
a = [1, 2, 3]
ERROR: LoadError: MethodError: no method matching resize!(::MyArray, ::Int64)
The function `resize!` exists, but no method is defined for this combination of argument types.

[KristofferC]

Naturally, one would expect that defining append!(::MyArray, xs) would enable push! as well.

I have a bit of a hard time seeing this. If anything it feels like the opposite.

[aplavin]

I personally expected both fallbacks to be honest :) Implemented append! for my array and indeed saw push! working.
Worked for quite some Julia versions, until today I ran tests on 1.11 and saw them broken.

[adienes]

one would expect that defining append!(::MyArray, xs) would enable push! as well

I can only speak for myself, but I would not have (and do not) expect this.

[oscardssmith]

Ah this is an interesting one. It should be pretty easy to change this so that it works again, but I'm not sure that this was intended. As a side note, the vararg push! does seem very odd...

[jakobnissen]

Perhaps related: #55459

FWIW, I don't see anything unexpected here. I interpret the docstring to mean that if both push and append is implemented, then append is equivalent to repeated pushing.