bad stack trace for errors in macros

[SimonDanisch]

macro test(x)
    if x.head == :(=)
        return :()
    elseif x.head == :(.)e # errors here
        return :()
    end
    return :()
end

begin



@test a = b
@test a.b


end

0.4:

ERROR: LoadError: MethodError: `*` has no method matching *(::Symbol, ::Irrational{:e})
Closest candidates are:
  *(::Any, ::Any, !Matched::Any, !Matched::Any...)
  *{T<:Number}(!Matched::Bool, ::T<:Number)
  *(!Matched::Complex{Bool}, ::Real)
  ...
 in include at ./boot.jl:261
 in include_from_node1 at ./loading.jl:304
 in process_options at ./client.jl:280
 in _start at ./client.jl:378
while loading /home/s/.julia/v0.5/FixedSizeDictionaries/test/runtests.jl, in expression starting on line 20

Note that this is the line number after the actual begin block.
0.5:

ERROR: LoadError: MethodError: no method matching *(::Symbol, ::Irrational{:e})
Closest candidates are:
  *(::Any, ::Any, !Matched::Any, !Matched::Any...)
  *{T<:Number}(!Matched::Bool, ::T<:Number)
  *(!Matched::Complex{Bool}, ::Real)
  ...
 in include(::ASCIIString) at ./boot.jl:240
 in include_from_node1(::UTF8String) at ./loading.jl:417
 in process_options(::Base.JLOptions) at ./client.jl:262
 in _start() at ./client.jl:318
while loading /home/s/.julia/v0.5/FixedSizeDictionaries/test/runtests.jl, in expression starting on line 11

it's a bit better, but still not that great. Especially, since even without the block, it still doesn't report the line number in the macro, and the macro can be arbitrarily deep in the stack trace but it will only report one line number...
julia --inline=no doesn't change anything.
Versions:

julia> versioninfo()
Julia Version 0.5.0-dev+3280
Commit b410b87* (2016-03-27 02:59 UTC)
Platform Info:
  System: Linux (x86_64-unknown-linux-gnu)
  CPU: Intel(R) Core(TM) i3-4130 CPU @ 3.40GHz
  WORD_SIZE: 64
  BLAS: libopenblas (USE64BITINT DYNAMIC_ARCH NO_AFFINITY Haswell)
  LAPACK: libopenblas64_
  LIBM: libopenlibm
  LLVM: libLLVM-3.7.1 (ORCJIT, haswell)

and

julia> versioninfo()
Julia Version 0.4.3
Commit a2f713d (2016-01-12 21:37 UTC)
Platform Info:
  System: Linux (x86_64-unknown-linux-gnu)
  CPU: Intel(R) Core(TM) i3-4130 CPU @ 3.40GHz
  WORD_SIZE: 64
  BLAS: libopenblas (USE64BITINT DYNAMIC_ARCH NO_AFFINITY Haswell)
  LAPACK: libopenblas64_
  LIBM: libopenlibm
  LLVM: libLLVM-3.3

@vtjnash saying:

it happens because the error doesn't get thrown directly, but instead gets unwound through flisp and is thrown directly by eval:

   424      else if (ex->head == error_sym || ex->head == jl_incomplete_sym) {
   425          if (nargs == 0)
   426              jl_error("malformed \"error\" expression");
   427          if (jl_is_byte_string(args[0]))
   428              jl_errorf("syntax: %s", jl_string_data(args[0]));
   429          jl_throw(args[0]); # <--- backtrace collected from here
   430      }

initially reported in #10595

[SimonDanisch]

Basically the same on windows for 0.4 and 0.5!