add setindex for named tuples

[matbesancon]

Given that Base.setindex is now documented, I think it makes some sense to have it defined for NamedTuples as well.
The implementation may be naive here and uses merge(::NamedTuple, itr), given that the left argument re-writes member already present in the first argument

[matbesancon]

One possible improvement is to re-use the code from merge directly:

function merge(a::NamedTuple{an}, b::NamedTuple{bn}) where {an, bn}
    if @generated
        names = merge_names(an, bn)
        types = merge_types(names, a, b)
        vals = Any[ :(getfield($(sym_in(n, bn) ? :b : :a), $(QuoteNode(n)))) for n in names ]
        :( NamedTuple{$names,$types}(($(vals...),)) )

[matbesancon]

cc @mbauman

[matbesancon]

I don't see how this creates a "test ambiguous" :/

[matbesancon]

re-opening for some failing checks

[matbesancon]

I added a bunch of tests in addition to the docstring

[matbesancon]

@mbauman I believe this is getting good. The failure is a connection problem

[mbauman]

I have two questions:

• Do we need two different setindex docstrings? Or can we define this in terms of a generic behavior that both ::Tuple and ::NamedTuple satisfy?
• For ::NamedTuple, should this also accept an integer index as well?

[matbesancon]

Do we need two different setindex docstrings? Or can we define this in terms of a generic behavior that both ::Tuple and ::NamedTuple satisfy?

In the case of NamedTuple, there isn't really a notion of out of bound for the symbol index. Otherwise we could merge the docstrings as you say, but do we need it?

For ::NamedTuple, should this also accept an integer index as well?

Is the syntax named_tuple[::Integer] really used much? It is not in the docstring of NamedTuple

[JeffBezanson]

I think this should support integer indices, but it has to be a separate method since it can't use merge directly.

[tkf]

• Do we need two different setindex docstrings? Or can we define this in terms of a generic behavior that both ::Tuple and ::NamedTuple satisfy?

I have an RFC on generic setindex specification: #33495. I think it covers Tuple and NamedTuple as well.