Meep geom

[HomerReid]

Replaces previous meepgeom PR.

Adds new folder libmeep_geom to top-level source tree with three working tests (pw-source-ll.cpp, bend-flux-ll.cpp, ring-ll.cpp`).

@oskooi @ChristopherHogan

[HomerReid]

In the most recent commit, I have

• renamed meep_geom to meepgeom

• removed bend-flux-ll from the test suite, as it was failing in the travis build. I wasn't able to debug exactly why this is, but I'm guessing it's because the code tries to write and then read back a largish HDF5 file which may not work in the travis build space. For now the test suite contains two tests which are passing in travis.

[stevengj]

Why does this need to be a function? Can't you just use a static initializer const material_type vacuum = {....}?

[HomerReid]

No, this is what was preventing the build from working on Travis. Previously I had all the initializations done the way you suggested, but these static structure initializations require the -std=c++11 compiler flag, which is not supported by the compiler versions used by Travis.

[stevengj]

Static structure initializations can be done even in 1990 ANSI C. You just can't use named fields. i.e.

const medium_struct vacuum_medium = {{1,1,1},{0,0,0},{1,1,1},{0,0,0}};
const material_data vacuum_material = {MEDIUM,NULL,NULL,&vacuum_medium};
const material_type vacuum = {(void*) &vacuum_material};

(Note that the unspecified extra fields in vacuum_medium are initialized to 0, per the C/C++ standard.)

[HomerReid]

OK, in the latest commit (a1117e6) I've eliminated the initialize_materials() function in favor of static structure initializers.

[stevengj]

I wouldn't bother with the the air synonym in the C++ interface.

[HomerReid]

OK, I removed it.