pmetrictest doesn't support ignore data point attribute values

[atoulme]

Component(s)

pkg/pdatatest

Is your feature request related to a problem? Please describe.

pmetrictest supports ignoring resource attribute values and metric attribute values, but does not support ignoring data point attribute values.

right now the workaround is to modify the data set with:

func ignoreMetricDataPointAttributes(t *testing.T, metrics pmetric.Metrics, attributeNames ...string) {
	for i := 0; i < metrics.ResourceMetrics().Len(); i++ {
		for j := 0; j < metrics.ResourceMetrics().At(i).ScopeMetrics().Len(); j++ {
			for k := 0; k < metrics.ResourceMetrics().At(i).ScopeMetrics().At(j).Metrics().Len(); k++ {
				metricToChange := metrics.ResourceMetrics().At(i).ScopeMetrics().At(j).Metrics().At(k)
				switch metricToChange.Type() {

				case pmetric.MetricTypeGauge:
					for l := 0; l < metricToChange.Gauge().DataPoints().Len(); l++ {
						dp := metricToChange.Gauge().DataPoints().At(l)
						dp.Attributes().Range(func(k string, v pcommon.Value) bool {
							for _, attributeName := range attributeNames {
								if k == attributeName {
									v.SetStr("*")
									break
								}
							}
							return true
						})
					}
				}
			}
		}
	}
}

Describe the solution you'd like

I would like to have the exact same option as offered by pmetrictest.IgnoreMetricAttributeValue

Describe alternatives you've considered

No response

Additional context

No response

[github-actions]

Pinging code owners:

• pkg/pdatatest: @djaglowski @fatsheep9146

See Adding Labels via Comments if you do not have permissions to add labels yourself.

[djaglowski]

Is the function just poorly named? If I'm not mistaken, metrics do not directly have attributes, so this function is actually operating data point attributes.

[atoulme]

Ah. It is. It also is surprising because it takes an attribute name and multiple metric names. Ideally I'd like both, but I can bear with it. I'll try it out and report back. Thank you for looking.

[djaglowski]

We can certainly look at changing it. There's likely a more intuitive signature.

[atoulme]

Going to close for now ; this is not a serious issue and while the API could be a bit clearer, I'm not sure it warrants a breaking change.