Additional Direction not needed

[sunpho84]

Solving this additional error, similar to #485

/usr/bin/ld: ./lib/libxchange.a(little_field_gather.o):/home/francesco/QCD/SORGENTI/tmLQCD/build_quda/xchange/../../xchange/little_field_gather.c:47: multiple definition of `Direction'; ./lib/libhmc.a(block.o):/home/francesco/QCD/SORGENTI/tmLQCD/build_quda/../block.c:78: first defined here

[kostrzewa]

In the current quda_work branch, there's only a single definition of this enum in block.c. Have you perhaps changed things locally?

[kostrzewa]

Ah, no, I see, there's another one in little_field_gather.c, but with a different numbering!

[kostrzewa]

block.c:

enum{
  NONE = 0, 
  T_UP = 1, 
  T_DN = 2, 
  X_UP = 3, 
  X_DN = 4, 
  Y_UP = 5, 
  Y_DN = 6, 
  Z_UP = 7, 
  Z_DN = 8
} Direction;                                                                                                                               

xchange/little_field_gather.c:

enum{
  T_UP = 0,
  T_DN = 1,
  X_UP = 2,
  X_DN = 3,
  Y_UP = 4,
  Y_DN = 5,
  Z_UP = 6,
  Z_DN = 7 
} Direction;                                                                                                                               

[kostrzewa]

NONE is never used apparently, but I honestly can't tell if this won't break things. It's not like we need a Lüscher-style deflated solver given the availability of DDalphaAMG. I thus don't much care if this breaks, but @urbach should decide since he spent so much time fixing this stuff...

[sunpho84]

It looks to me like the Direction structure in block.c is simply dead code, as it is not referred anywhere in the TU, so I would be tempted to say that removing it should have no impact.

Indeed both are contained in a .c file, and block.c compiles no matter the presence of the structure. Am I too naive...?

[kostrzewa]

I will just merge this in now. If it breaks things, somebody else will have to fix it...

[sunpho84]

I'll keep my fingers crossed...