malloy-gotchas-queries

Use the `having:` clause to filter on aggregates after aggregation, and reserve `where:` for filtering rows before aggregation on dimensions or raw columns. Writing `where: n > 10` on a measure produces an "Aggregate expressions not allowed in where" compile error.

Malloy's `order_by:` resolves only against the query's output columns, not source fields. When grouping by a joined field, first alias it in `group_by:` (e.g., `group_by: yr is races.season_year`) and then reference the alias in `order_by: yr`.

Method syntax is required for aggregates over a joined path, such as `inventory_items.item_cost.sum()`; the plain `sum(joined.field)` form fails to compile. Scalar functions like `round`, `floor`, and `ceil` never use method syntax and must be called as `round(x, 2)`.

`ts.month` truncates a timestamp to the start of the month and returns a timestamp, which is correct for time series charts. `month(ts)` extracts the month as an integer from 1 to 12, which suits cross-year comparisons.

Charts render only the first aggregate in a view, so a `# bar_chart` or `# line_chart` view must contain exactly one aggregate. For multiple metrics, nest separate chart views inside a `# dashboard` or use `y=['revenue','cost']` for multi-measure series.