Conform closer to OVF 2.0 format specifications

GPMueller commented 6 years ago

The requirements are:

a rectilinear lattice with a single-atom basis cell should work identically between atomistic and micromagnetic
specification-conformant files should always be read in correctly
a conformant parser (e.g. micromagnetic code) should be able to read our non-conformant files

The latter two points would mean non-conformance would have to be a pure extension of the format, without changing keywords or their meaning.

One concern for me is that OVF keeps everything in x, y, z representation with one node representing one data point, whereas we may want to generalize it to arbitrary bravais lattices with basis cells with potentially multiple nodes per cell. Otherwise, we would have to represent such cases as irregular meshes...

Another concern is that the valuedim seems to specify additional dimensions (the specs are unclear on this), meaning that 3 columns of data always represent the positions and the remaining columns represent the actual data. This seems very wasteful - maybe it is only supposed to be done this way for meshtype irregular? Otherwise I suggest non-conformance.

GPMueller commented 5 years ago

Improved specification in Readme on feature-pegtl, see https://github.com/spirit-code/ovf/tree/feature-pegtl#file-format-v20-specification-

GPMueller commented 5 years ago

Significantly improved the specifications written in the Readme and the adherence of the library to the specifications, merged into develop with 55c326c9a7b11d999a05f98c2e095188527eb33b.

spirit-code / ovf

Conform closer to OVF 2.0 format specifications #8