Bogdan Drutu

> @bogdandrutu just to clarify, are you saying https://github.com/open-telemetry/opentelemetry-specification/pull/2617 should be reverted? That PR split process.disk.io and process.network.io by eliminating the direction attribute. Do you think both of this metric...

@reyang @tigrannajaryan one thing that definitely this mistake open a huge cane of warms, see https://github.com/open-telemetry/opentelemetry-specification/pull/2706 :) Should that be an attribute or not? When do we apply the "exception"....

> That being said, you seem to imply approvers do not review the actual content, including TC members, so please be more vocal about it, so we avoid misunderstandings My...

@lquerel I don't agree with the statement that format influenced any of the split decision. The decision to split was mostly because of "total being or not being useful" or...

@lquerel great explanation, I think we are on the same page, but also I think that they proposed change (that I am calling as mistake) is moving in the opposite...

I am +1 on this

@carlosalberto some questions: > Agree to no longer use the deprecated delimiter in proto, as this breaks JSON. Is this the rename with a "deprecated_" prefix? Proto also has an...

@Aaronontheweb why the current "UpDownCounter" is not good? Curios what exactly is the metric about the actors that you want to report.

> if the CRD is present then it is denoted by 1 as the metric value and if it's not (if it's deleted) then it is denoted by 0 ...so...

@Aaronontheweb "UpDownCounter" is not for "rate". The "UpDownCounter" is as a name says a counter (sum) that supports positive and negative values, so your backlog will do "+1" when element...