New test "correctly print values to error messages" fails on Windows

[tkelman]

This recently added test is failing on Windows.

The output of expand(parse("\"a\"=1")) with the most recent Windows prerelease binary or my own freshly-built Julia is :($(Expr(:error, "invalid assignment location \"alue\""))). On the 0.2.0 release, it is :($(Expr(:error, "invalid assignment location #<julia_value>"))).

[tkelman]

Does not appear to have fixed the problem.

               _
   _       _ _(_)_     |  A fresh approach to technical computing
  (_)     | (_) (_)    |  Documentation: http://docs.julialang.org
   _ _   _| |_  __ _   |  Type "help()" to list help topics
  | | | | | | |/ _` |  |
  | | |_| | | | (_| |  |  Version 0.3.0-prerelease+1808 (2014-03-02 17:08 UTC)
 _/ |\__'_|_|_|\__'_|  |  Commit 73d6a45* (0 days old master)
|__/                   |  x86_64-w64-mingw32

julia> expand(parse("\"a\"=1"))
:($(Expr(:error, "invalid assignment location \"alue\"")))

[JeffBezanson]

Thanks for checking.

@vtjnash the jl_static_print hack in print.c seems not to be working on windows. Is this expected? If so I will just work around it.

[vtjnash]

There shouldn't be any difference in behavior

[JeffBezanson]

It seems to be falling back to printing #<julia_value>, which is where the "alue" string comes from. AFAICT this would happen if it can't look up the address of jl_static_print. No idea why.

[vtjnash]

Ah yes. It's much harder to get the dynamic linker to find functions on windows since you have to tell it exactly where to look

[vtjnash]

Actually, on windows, we don't even attempt to look since RTLD_SELF and RTLD_DEFAULT are not defined (and we wouldn't find the jl_static_show at those locations anyways). Since those don't exist, there is no possible way to find the symbol (unless flisp is statically linked to libjulia, in which case, we can query our own address space with GetModuleHandleExA like we do in jl_load_dynamic_library_)

[JeffBezanson]

flisp is statically linked inside libjulia (libflisp.a).

[tkelman]

2-years-old bump. This still gives

julia> expand(parse("\"a\"=1"))
:($(Expr(:error, "invalid assignment location \"#<julia_value>\"")))

on Windows. Should we be able to do better with libflisp linked into libjulia? This was marked 0.4.x...