Writing a macro generates code utilizing NamedTuple whose value is calculated

[runapp]

I'm writing a macro tha packs it's parameter and @FILE into a NamedTuple and push it into a Vector, like this:

testvec=Vector{NamedTuple{(:f,:s),Tuple{Int,String}}}()
# push!(testvec,(f=0,s="abc"))
macro testm(a)
    return quote
        b=a+1
        push!(testvec,$(esc(
            (f=b,s=@__FILE__)
        )))
    end
end
@testm 2

My think is to try avoiding other ways than esc to prevent f and s from being renamed, or to make b work in escaped environment. However I tried $(esc(b))=a+1 and other escape forms, with none of them succeeded.

Any suggestions?

[KristofferC]

Please use https://discourse.julialang.org/ for general Julia questions.

[runapp]

Forget about it.
Finally got $(esc(begin b=a+1 end)) to work.
I post it here rather than asking on discourse or StackOverflow only to see if anybody wants some new syntax to make it more elegent, such as makes global accept symbols.

[JeffBezanson]

I think there is also a hygiene bug here though:

julia> macro id(x)
       x
       end
@id (macro with 1 method)

julia> @id (a=1,)
(#12#a = 1,)

[tkluck]

@KristofferC I'm seeing this behaviour on Julia master, also related to named tuple hygiene:

$ julia
               _
   _       _ _(_)_     |  Documentation: https://docs.julialang.org
  (_)     | (_) (_)    |
   _ _   _| |_  __ _   |  Type "?" for help, "]?" for Pkg help.
  | | | | | | |/ _` |  |
  | | |_| | | | (_| |  |  Version 1.4.0-DEV.189 (2019-09-21)
 _/ |\__'_|_|_|\__'_|  |  Commit 36220ed9c7* (3 days old master)
|__/                   |

julia> macro named_tuple_hygiene(key, value); return :( (a=1, $key => $value) ); end
@named_tuple_hygiene (macro with 1 method)

julia> @named_tuple_hygiene(:b, 2)
ERROR: syntax: invalid named tuple element "Main.=>(:b, 2)"
Stacktrace:
 [1] top-level scope at REPL[2]:1

I think the root cause is that named tuple syntax Angers The Lisp Gods: within a Expr(:tuple, ...), the expression Expr(:call, :(=>), key, value) has a meaning different from "call => on key and value".

One proper fix would be to parse it differently: e.g. we should have (a = b, c => d) parse to something like

Expr(:tuple, Expr(:(=), a, b), Expr(:(=>), c, d))

An alternative fix is to allow pair-valued expressions as elements of named tuples (currently a syntax error), e.g.

julia> keyval = :b => 2
:b => 2

julia> (a=1, keyval)
ERROR: syntax: invalid named tuple element "keyval"
Stacktrace:
 [1] top-level scope at REPL[4]:1

But given that you've worked on this recently, maybe we can come up with a solution that only touches the macro hygiene logic. What do you think?

[KristofferC]

I'm not sure why I was pinged here?

[tkluck]

I pinged you because of this:

Is there someone else who is in a better position to check it out?

[KristofferC]

Oh, that's just cause I backported that commit to some release. #32464 is the original PR.

[tkluck]

Ah that makes sense, sorry for the noise! Added a comment there as well.