interface interfaceName: string, [fieldBinding: expression]+

Map fields of this dataset to a pre-defined interface.

For each well-defined interface, some number of field:type mappings are
defined, which makes the dataset work according to the needs of that interface.
The most common such interface is "metric" which makes Observe calculate
summaries of the metrics, and present them in the metric browser. Each named
field in the interface is also referred to as a interface field name because
the name identifies a particular role for the field (for example "this field is
the name of this metric.")

The "metric" interface requires the following fields:

If a column of the expected name and type already exists, it will be picked up
as the "right" column, whereas if the intended column has another name, you
will need to specify it:

interface "metric", metric:my_name, value:the_reading

This interface can currently bind these interface field names to fields with
different names, without renaming the columns, such as shown in one of the
examples below. For example, there may be a field named "my_name" that is bound
to the "metric" interface field name because it contains the metric name in
this dataset. Because using a differently-named column for the particular
interface field can be confusing, Observe strongly recommends making the name
of the field be the same as the name of the interface field -- use rename_col metric:my_name before declaring interface "metric".

Some metric data sources may provide useful metadata. If the metric type, unit size,
and description are available or can be defined in OPAL, you can add them to the
metrics interface. When these options are defined, Observe can find the metric
metadata in the columns for these optional string fields.

Note that interface is transitive, meaning child datasets will inherit the
interface from their parents. This makes datasets available for exploration
without requiring an explicit declaration on every dataset definition.

Categories

• Metadata

Accelerable

interface is always accelerable if the input is accelerable. A dataset that only uses accelerable verbs can be accelerated, making
queries on the dataset respond faster.

Examples

Make this dataset implement the 'metric' interface, after first extracting the metric
name and metric value from the FIELDS object. Because the columns have names matching
the interface needs, they don't need to be named again.

Make this dataset implement the 'metric' interface. Bind the existing column containing metric names ('metricNameColumn') to the 'metric' interfaceName, and the existing column containing 'float64' metric values ('metricValueColumn') to the 'value' interfaceName.

Make this dataset implement the 'log' interface, after first extracting the
log message from the FIELDS object. Because the column name matches the
interface expectation of "log", it does not need to be named again.