Pivot/Table “Absolute change” (time comparison delta) ignores new dimension members (current period has value, previous period missing) and shows 0 change

[paul-villalobos]

When using a Pivot/Table in Rill with comparison_mode: time (YoY comparison), the automatically computed Absolute change (delta) does not include contributions from new dimension members that have sales in the current period but no rows in the comparison period (e.g., new distributors this year). These members appear with a delta of 0, even though their full current-period value should contribute to the absolute change.

This causes totals at higher levels (e.g., Region) to understate the absolute change because the “new” members’ sales are not counted in the delta.

To Reproduce

Steps to reproduce the behavior:

Open a Canvas dashboard that has:
    defaults.comparison_mode: time
    a Pivot (or Table) grouped by a dimension that can have new members (e.g., Distribuidora)
Set the time range to a period where some members exist only in the current period (e.g., YTD this year) and not in the previous year.
In the Pivot, group rows by:
    Region
    Distribuidora and include a measure like venta (sales).
Enable/show the automatically computed comparison columns (e.g., previous, absolute change / delta, percent change).
Observe that distributors that are new this year (have venta > 0 in current period and no data in previous period) show delta = 0 and do not contribute to the Region-level absolute change.

Example from our case:

Region “Norte / Oriente” current sales ≈ 916,007
Previous year sales ≈ 720,000
Several new distributors (e.g., Disedima, CSM, ICO Piura, OCOSAC, San Juan) have current sales but no previous-year rows
The Pivot’s automatic delta does not reflect their contribution (shows 0 for those rows and understates the Region delta)

Expected behavior

For any dimension member where the comparison period has no data:

Treat the previous value as 0 (not “missing”)
Therefore:
    absolute change (delta) = current − 0 = current
    percent change can be NULL/undefined (division by zero), but delta should still be correct

At aggregated levels (e.g., Region), the absolute change should equal:

(sum of current period values) − (sum of previous period values) including the full contribution of new members.

Desktop (please complete the following information):

OS: [e.g. Windows 11 / macOS 14]
Browser: [e.g. Chrome]
Version: [e.g. 125]

Additional context

This happens in a Canvas dashboard Pivot using a metrics view (semantic layer) where the base measure is a standard sum (e.g., SUM(venta)).
Workaround: define an explicit YoY absolute change measure in the metrics view (e.g., venta_yoy_abs with COALESCE(previous, 0)) and display that measure instead of relying on the Pivot’s automatic delta. However, the expectation is that Rill’s built-in “absolute change” for time comparisons should handle missing previous-period members as 0 for correct deltas.