In order to support #7 (without a multi-process KV store), and as an alternative fix for #27, and perhaps to improve storage space, there should be a file-based cache implementation.
In this schema, one would keep the CDB library for the meta-data and other information, but all the scriptlet contents could be stored on disk as individual files. (This feature could also be used for scriptlets that require prologues, currently Bash and Python.)
In order to support #7 (without a multi-process KV store), and as an alternative fix for #27, and perhaps to improve storage space, there should be a file-based cache implementation.
In this schema, one would keep the CDB library for the meta-data and other information, but all the scriptlet contents could be stored on disk as individual files. (This feature could also be used for scriptlets that require prologues, currently Bash and Python.)