v0.60.0

Breaking

• Renaming xan from --path to --root.
• Renaming xan separate -T/--txt to -L/--lines.

Features

• Adding xan count -c/--check-alignment & -H/--human-readable.
• Adding xan sort -e -z/--compress.
• Adding xan sample -g -S/--sorted.
• Adding xan from -f=(json|ndjson) --model.
• Adding xan cat rows -I/--intersection & -U/--union.
• Adding xan top -g -S/--sorted.
• Adding the sort, dedup, flatten & flat_map moonblade function.
• Adding support for list & map columns when using xan from -f=parquet.

Fixes

• Fixing text wrapping across tool, especially xan flatten -w & xan flatten -F.
• Fixing xan select -e & xan map plural clause flattening given list.
• Fixing xan from -f=parquet not emitting correct column names for list & map columns.

Performance

• Introducing a fast path for xan sort -e when input fits in a single chunk.
• Faster xan from -f=ndjson & -f=parquet.
• Amortizing xan sample allocations.
• Improving performance of xan freq & xan p freq.

Quality of Life

• Better record size estimation for xan sort -e.
• Better xan flatten -c when string width cannot be computed correctly (because of emojis typically).