We already have .gz and .lz4 compression via spork utilities, so dealing with streaming-compression of our textual inputs should be trivial. We also have the ability to compute delta-diffed serialized histories from M4. Both of those provide a more efficient storage medium than our current bloated text input and text output. Recommend shifting to a compressed format [or supporting the option] for both, with the ability to coerce compressed outputs to plaintext if necessary.
Could also examine binary outputs if we use static schemas...
We already have .gz and .lz4 compression via spork utilities, so dealing with streaming-compression of our textual inputs should be trivial. We also have the ability to compute delta-diffed serialized histories from M4. Both of those provide a more efficient storage medium than our current bloated text input and text output. Recommend shifting to a compressed format [or supporting the option] for both, with the ability to coerce compressed outputs to plaintext if necessary.
Could also examine binary outputs if we use static schemas...