mirror of
https://github.com/wassname/catalyst.git
synced 2026-07-01 02:29:54 +08:00
Eddie Hebert
b450ab841f
Fix behavior in minute mode history with frequency `1d`, where on the day immediately following an adjustment action, the overnight adjustment would not apply. (However the adjustment would be applied after a 1 day lag.) The root cause of the bug was that the history data for minute mode when using `1d` stitches together a sliding window of the daily data for previous and the current minute. That daily data sliding window and corresponding adjustments was being read as if the data was being viewed from on the last day of the window; however in this case the data is being viewed from the day after the window has completed. The difference in view points requires the adjustments to popped and applied by the adjusted array one index earlier. The fix uses the `extra_slot` value as signifier on whether the data is being viewed on the following day and then accordingly adjusts the index of the mulitpy object. Also, change the split and merger test data ratios to have different values, to ensure that different adjustment values are applied; as opposed to doubling up on just one of the values.