Right now osrm-datastore needs to know from which file it can load the data that goes into a certain named block in memory. For example, for the street names we explicitly load the data from the file .osrm.names here. We already automated the block discovery, but to unlock loading from arbitrary files we would need to build an index, that maps storage block identifier, like /common/names/blocks, to a file path data.osrm.something.

The use-case that we would want to support here is being able to combine the output of multiple toolchain steps to a single archive:

Combining for easy handling and deployment

cat data.osrm.* | gzip - > data.osrm.gz

Extraction for loading to memory

cat data.osrm.gz | gunzip - > data.osrm
./osrm-datastore data.osrm