steed 0.1.1

A drop-in replacement for tee with file buffering and output queue management.

Pipelines may wish to use tee to cache intermediate steps as a failsafe in case of a downstream failure. This strategy is limited by the fact that the speed at which tee writes to disk is bottlenecked by the speed at which the downstream pipeline reads from its stdout. Therefore tee cannot finish writing to disk before its input has been completely output by the downstream pipeline. An option is to store all input from stdin in an in-memory buffer, but this can lead to out-of-memory failures. The steed command overcomes this limitation by writing to the on-disk file continuously, while simultaneously reading from it to obtain a size-limited buffer queue that is output to stdout. This allows steed to complete writing to disk before it has finished outputting to downstream commands, improving its utility as a cacheing mechanism.