Some objects (location being the obvious one) are likely to be repeated multiple times within a dataset. To prevent this duplication we should perhaps permit a cache key on the same level as metadata and records that will contain an object with keys (md5, sha?) that can be referred to later in the file. Eg
"cache": { "e54789a3": {a GeoJSON boject},
"680f038d": "A base64 encoded file"}
}
Records in the cache could then be referred to in other elements of the file with a cache object
"location":{"ref$": "e54789a3" }
Some objects (location being the obvious one) are likely to be repeated multiple times within a dataset. To prevent this duplication we should perhaps permit a cache key on the same level as metadata and records that will contain an object with keys (md5, sha?) that can be referred to later in the file. Eg
Records in the cache could then be referred to in other elements of the file with a cache object