@inbounds overrides val variable in macros. Wrong usage of local

[0x0f0f0f]

It took me a while when debugging a compiled pattern symbolics matcher to realize why pattern variables i've named val were overridden by other values. I've ended up adding a prefix to those symbols to avoid the issue. The macro shouldn't be escaped, as fresh symbols are needed to avoid clashes.

julia> macro foo(x,i)
           v = [1,2,3,4,5]
           quote
               val = $x
               t = @inbounds ($v)[$i]
               println(val)
           end
       end

julia> @foo(2,3)
3

The reason is because

macro inbounds(blk)
    return Expr(:block,
        Expr(:inbounds, true),
        Expr(:local, Expr(:(=), :val, esc(blk))),
        Expr(:inbounds, :pop),
        :val)
end

My intuition was always that nested macro calls that "access" the same symbol, should hygiene the symbol in the innermost macro expansion, but it seems it's not the case. What's the correct pattern of macro hygiene for these cases? shall I file a PR renaming that val in @inbounds to something less likely to clash with outer identifiers?

julia> @macroexpand @foo(2,3)
quote
    #= REPL[39]:4 =#
    var"#9332#val" = 2
    #= REPL[39]:5 =#
    var"#9333#t" = begin
            $(Expr(:inbounds, true))
            local var"#9332#val" = ([1, 2, 3, 4, 5])[3]
            $(Expr(:inbounds, :pop))
            var"#9332#val"
        end
    #= REPL[39]:6 =#
    Main.println(var"#9332#val")
end

julia> versioninfo()
Julia Version 1.10.2
Commit bd47eca2c8a (2024-03-01 10:14 UTC)
Build Info:
  Official https://julialang.org/ release
Platform Info:
  OS: Linux (x86_64-linux-gnu)
  CPU: 8 × 11th Gen Intel(R) Core(TM) i7-1165G7 @ 2.80GHz
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-15.0.7 (ORCJIT, tigerlake)
Threads: 1 default, 0 interactive, 1 GC (on 8 virtual cores)
Environment:
  JULIA_EDITOR = code
  JULIA_NUM_THREADS = 

Installed via juliaup

[vtjnash]

This sounds like a bug nearly identical to #53667, but yeah we could rename that to something more awkward to avoid clashes while this bug is still pending in the hygiene computation

[Pangoraw]

Linking #48910 which has a similar cause/problem (although with @inline).

[0x0f0f0f]

I'm confused by the behavior of local?

julia> begin 
           x = 2 
           begin 
               local x = 3
           end
           x
       end
3

julia> x = 2 
           begin 
               local x = 3
           end
           x
2 # works correctly in global scope? 

julia> begin 
           x = 2 
           if x > 1
               local x = 3
           end
           x
       end
3

julia> begin 
           x = 2 
           for i in 1:x
               local x = 3
           end
           x
       end
2

I'm referring to the manual page on scoping https://docs.julialang.org/en/v1/manual/variables-and-scoping/

Notably missing from this table are begin blocks and if blocks which do not introduce new scopes. The three types of scopes follow somewhat different rules which will be explained below.

If local has no effect on begin and if blocks, why use it in @inbounds and @inline?

[giordano]

Also similar to #41451 and #48015 (for logging macros).