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
We are still stuck with v0.0.4. With maturing the exposition format, we should bump the version to utlimately 1.0.0. Also, we need to figure out how to document older versions, and what requirements we have for support of various versions by implementers.
The way I'm thinking is that a 1.0.0 parser would also handle 0.0.4 output, with no more than minor mangling (e.g. the comment escaping). The only real problem with that is if we introduce the end marker.
On a different note, it'd be helpful to include information about Prometheus in the Content-Type header value. text/plain; version=0.0.4 doesn't help to identify the format definition if someone doesn't know about Prometheus.
That brings up a cyclic point... It would be best to refer to the RFC in
there, but we can't do that for the year or so it takes to get to the
RFC... A backlink we commit to hold accessible forevermore which can then
point to the specs?
A possible way out would be to specify "MUST link to $url OR specify RFC
$tbd"
Richard
Sent by mobile; excuse my brevity.
Activity
beorn7 commentedon Sep 9, 2016
We should also get clarity where and how to advertise/mark the version, cf. prometheus/common#34
brian-brazil commentedon Sep 10, 2016
The way I'm thinking is that a 1.0.0 parser would also handle 0.0.4 output, with no more than minor mangling (e.g. the comment escaping). The only real problem with that is if we introduce the end marker.
grobie commentedon Sep 10, 2016
On a different note, it'd be helpful to include information about Prometheus in the Content-Type header value.
text/plain; version=0.0.4doesn't help to identify the format definition if someone doesn't know about Prometheus.RichiH commentedon Sep 11, 2016
tools: update the code block in PD Control (prometheus#545)
hdost commentedon Feb 3, 2022
https://github.com/OpenObservability/OpenMetrics seems to make this irrelevant.
lasomethingsomething commentedon Nov 18, 2026
@jan--f Any reason not to close?