Currently in the instrumentation, we ensure that we use maps and have unique keys ALWAYS. However, when parsing, the data might come from an arbitrary source or some languages might not have facilities for maps and say for attributes that might arrive on the wire and be appended or for lists, we should define what to do when we encounter duplicated keys with differing values.
For example
1. As list
/*foo='bar',foo='baz',foo='fizz'*/
which we should then interprete as a list
key: foo
values: ['bar', 'baz', 'fizz']
or perhaps we can be stricter and say the last value wins so
2. As set
/*foo='bar',foo='baz',foo='fizz'*/
which we should then interprete as a list
/cc @leepengxiang
Currently in the instrumentation, we ensure that we use maps and have unique keys ALWAYS. However, when parsing, the data might come from an arbitrary source or some languages might not have facilities for maps and say for attributes that might arrive on the wire and be appended or for lists, we should define what to do when we encounter duplicated keys with differing values.
For example
1. As list
which we should then interprete as a list
or perhaps we can be stricter and say the last value wins so
2. As set
which we should then interprete as a list
key: foo value: 'fizz'/cc @leepengxiang