Tags: liquidaty/zsv
Tags
Sql dup colname bug (#631) feat: handle duplicate input column names across sql, sheet, and flatten A repeated header name previously crashed zsv sql (double-free/SIGABRT); it is now handled gracefully across commands: - sql: fix the double-free; add --rename-duplicate-columns to disambiguate repeats (a,b,a -> a,b,a_2); on a strict failure, emit an actionable parse-time error naming the column, its positions, and the remedy flag; note each rename on stderr. - sheet: sqlfilter and pivot disambiguate duplicate columns unconditionally (no flag surface) so such files stay filterable/viewable. - flatten: ensure unique output column names by default, noting renames on stderr. Build: add .d dependency tracking for the sheet unity build (edits to #included sources now trigger a rebuild); touch both utf8proc extraction outputs so the multi-target rule runs once. Tests: dupcol coverage for sql (crash/rename/error), flatten, and an interactive sheet sqlfilter test. Dedupe paths verified under AddressSanitizer. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add fast parser (#563) * Performance optimizations * Faster scalar engine * Add `--parser fast` option to use faster CSV scanner for standard-quoted CSV * Available for NEON, SSE2 and AVX * More robust tests - expanded test coverate - tmux tests more stable * More robust benchmarking - verify correctness - broader set of inputs - add qsv & xan - save platform-based results markdown * Add make test-asan target * Stabilize overwrite cleanup routines Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
PreviousNext