You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
one thing that I haven’t implemented yet is the network.protocol.name opt-in attribute — right now we’re exporting it all the time
we’ve briefly talked about it in the last SIG meeting, I’m slightly opposed to removing it — to me the network.protocol.(name|version) make a pair, and sending just the version seems incomplete
and also we have some legacy HTTP clients that still technically support the deprecated protocols
network.protocol.name was changed from recommended to opt-in because:
removing the default value (http) would result in a high noise to value attribute stamped on all HTTP spans
I think we may be able to address this by making network.protocol.name conditionally required when network.protocol.version is provided, and making http the default value (only) when network.protocol.version is provided.
The text was updated successfully, but these errors were encountered:
From @mateuszrzeszutek
network.protocol.name
was changed from recommended to opt-in because:http
) on a non-required attribute made it ambiguous whether the instrumentation just doesn't capture the attribute or whether it does capture it but it was the default value (Attributes should only have default values if supplying the attribute is (otherwise) required #385)http
) would result in a high noise to value attribute stamped on all HTTP spansI think we may be able to address this by making
network.protocol.name
conditionally required whennetwork.protocol.version
is provided, and makinghttp
the default value (only) whennetwork.protocol.version
is provided.The text was updated successfully, but these errors were encountered: