Testing and feedback welcomed, heading in to 1.0 release soon.
Answering a FAQ:
DuckDB is probably the closest analogue, yes.
Unlike duck,
(1) sqawk preserves unchanged CSV rows, byte for byte
(2) sqawk uses sqlite-like dynamic typing, which is great for heterogeneous, messy Unix text files
(3) simpler, more flexible invocation model that does not require referencing file paths inside the query
(4) no spilling externally, which can make things faster
Duck is better if the working set is larger than Ram, since it spills, and where columnar batches are heavy.
Answering a FAQ: DuckDB is probably the closest analogue, yes.
Unlike duck, (1) sqawk preserves unchanged CSV rows, byte for byte (2) sqawk uses sqlite-like dynamic typing, which is great for heterogeneous, messy Unix text files (3) simpler, more flexible invocation model that does not require referencing file paths inside the query (4) no spilling externally, which can make things faster
Duck is better if the working set is larger than Ram, since it spills, and where columnar batches are heavy.