Paul Trowbridge 2ee0d18f2e Fix large dataset loading in Forecast view ... - Switch server Arrow encoding from tableFromJSON (row objects) to tableFromArrays (column arrays) — cuts peak Node heap 3-5x for large datasets by avoiding one JS object per row - Remove unused pf.log JOIN from data endpoint; forecast rows only - Load Perspective viewer with direct table reference instead of worker Server object — fixes "No Table attached" error on large datasets where named-table registry lookup raced against WASM initialization - Pre-emptively clean up stale named table in worker registry before creating, eliminating the "already exists" retry path that silently swallowed errors (finally ran but flash never fired) - Strip cfg.table from restore configs since table is loaded by reference - Throttle progress bar updates to 100ms intervals (was every chunk) - Persist load errors until dismissed; add console.error for devtools Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>		2026-05-21 20:52:47 -04:00
..
log.js	Undo deletes rows in place instead of reloading the forecast	2026-04-28 20:21:13 -04:00
operations.js	Fix large dataset loading in Forecast view	2026-05-21 20:52:47 -04:00
sources.js	Add per-source default Perspective layout	2026-04-29 22:31:23 -04:00
tables.js	Expose pf_note/pf_op in forecast data; fix tables list duplicates	2026-04-28 19:51:45 -04:00
versions.js	Refactor sources UI, rename pf_ system cols, replace filter builder with raw SQL	2026-04-03 00:47:57 -04:00