Add tests for attribute JSON marshalling

[dashpole]

Forked from #7175 (comment)

This adds a test for JSON marshaling of attribute sets.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 82.9%. Comparing base (9798759) to head (e6c5f5c).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@          Coverage Diff          @@
##            main   #7268   +/-   ##
=====================================
  Coverage   82.9%   82.9%           
=====================================
  Files        265     265           
  Lines      24894   24894           
=====================================
+ Hits       20639   20646    +7     
+ Misses      3879    3873    -6     
+ Partials     376     375    -1     

see 3 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[MrAlias]

Thanks for the added tests, this is a good addition to the package.

The original comment inspiring this was about the unmarshlilng of this data. Do we need added checks for unmarshling, or is that just not a possible operation?

[dashpole]

Currently, unmarshaling our attribute.Value type doesn't work properly. When I tried, it correctly unmarshaled the list and the key of each element, but the value was always invalid. I think we could implement UnmarshalJSON for the Value type to make that round-trip work properly, but it doesn't today. I figured for the purposes of unblocking #7175 it would be enough to test the MarshalJSON output.

See

opentelemetry-go/attribute/value.go

Lines 262 to 270 in be1e57f

	func (v Value) MarshalJSON() ([]byte, error) {
	var jsonVal struct {
	Type string
	Value any
	}
	jsonVal.Type = v.Type().String()
	jsonVal.Value = v.AsInterface()
	return json.Marshal(jsonVal)
	}

. We have some custom logic for marshaling, but nothing for unmarshaling.

[MrAlias]

So if we add another hash field to the Set will the unmarshal behavior remain the same (produce invalid values), or will it start to error?

[dashpole]

Ah, you can't unmarshal into a Set today, but you can unmarshal into a []KeyValue. The former still won't work, and the latter will still work since we aren't changing KeyValue at all.