Rules / Naming Conventions

Partition name should match table name for single partition tables

info PARTITION_NAME_SHOULD_MATCH_TABLE_NAME_FOR_SINGLE_PARTITION_TABLES · ported · model layer · scope: Table

What it checks

Regular tables with exactly one partition whose name differs from the table name. Calculated tables and calculation groups are not checked.

Each finding names the table, as 'Sales', not the partition whose name differs, so open the table in the TMDL file to see the name the partition is carrying.

Example

Fires the rule

table Sales
	column 'Order ID'
		dataType: int64
		sourceColumn: OrderID
	column Amount
		dataType: decimal
		sourceColumn: Amount
	partition SalesData = m
		mode: import
		source = Sql.Database("contoso", "Sales")

After the fix

table Sales
	column 'Order ID'
		dataType: int64
		sourceColumn: OrderID
	column Amount
		dataType: decimal
		sourceColumn: Amount
	partition Sales = m
		mode: import
		source = Sql.Database("contoso", "Sales")

Why it matters

A single-partition table has no reason for its partition to carry a different name, and when it does it is usually the table's old name from before a rename. Refresh logs, error messages, and the TMDL file all name the partition, so a mismatch sends the reader looking for a table that no longer exists.

How to fix it

The partition's name sits on its declaration line, so partition SalesData = m under table Sales becomes partition Sales = m. Nothing else in the TMDL file refers to a partition by name, so on a single-partition table the rename changes nothing but the label in the refresh log. Power BI Desktop has no partition list and no field for a partition name: it names the partition when it creates the table and leaves the name alone afterwards, so a mismatch on a Desktop project points at a hand edit, a table renamed outside Desktop, or a model migrated from another tool. If a whole model needs correcting, Tabular Editor's partition editor renames partitions without opening each file.

When to ignore it

A partition that the people who run the refresh already know by another name can keep it, for example one named after the stored procedure that fills it, because that is the name they will look for when the refresh fails. Outside that, a single partition named anything but its table is a leftover, and the rename costs nothing.

To ignore this rule on one object, add annotation pbiplint.ignore = PARTITION_NAME_SHOULD_MATCH_TABLE_NAME_FOR_SINGLE_PARTITION_TABLES under the object in its TMDL file. Power BI Desktop keeps the annotation. To turn the rule off for a whole project, set "PARTITION_NAME_SHOULD_MATCH_TABLE_NAME_FOR_SINGLE_PARTITION_TABLES": "off" under rules in pbiplint.config.json.

Quirks

The condition tests tables with exactly one partition. A table with none, and a table with several, is never reported, whatever its partitions are called.
The rule also needs every part of a table's declaration, which TMDL lets sit in more than one file (Power BI Desktop writes each table in one). While pbiplint could not open a model file or folder, or a parse issue took a line that could be a table line, such as a misspelt table, the rule reports nothing, because a part of the table in what pbiplint missed could hold another partition, or a calculated one, and pbiplint does not guess what a file it could not read says. A parse issue inside a declaration, such as a property indented with spaces, does not stop the rule. The skipped line gives the reason, a model file could not be fully read, and a notice names what pbiplint could not open, or the file's own PARSE_ISSUE finding names the line.

LARGE_TABLES_SHOULD_BE_PARTITIONED asks a large table for more than one partition, and a table that takes its advice leaves this rule's condition.
TRIM_OBJECT_NAMES reads partition names too, so a partition renamed with a stray space is reported by both rules: the names still do not match.

Ported from Microsoft's Best Practice Analyzer ruleset (opens in a new tab).

Check a model for this Improve this page (opens in a new tab)