Consensus: Label values are strings only; not int, not float, not bool

[RichiH]

No description provided.

[curlup]

How about histogram bounds in labels? (https://github.com/RichiH/OpenMetrics/issues/)
Say, the upper bound label is le="0.01" which is string representation of a float value, and I suppose a user might want to select them on le<1 or alike condition.

[brian-brazil]

That's a matter for the text representation, and the ingesting backends to handle. I presume/hope we'll continue to use the le label as we currently do.

[jkschneider]

le implies that histograms are always cumulative. But this is the opinion of Prometheus. Do we intend to impose that? Atlas uses normal histograms.

[brian-brazil]

Yes, this is a hard requirement. Converting between cumulative and non-cumulative is not hard. Internally for example most Prometheus clients use non-cumulative histograms for contention reasons.

…

On Sat 11 Aug 2018, 23:28 Jon Schneider, ***@***.***> wrote: le implies that histograms are always cumulative. But this is the opinion of Prometheus. Do we intend to impose that? Atlas uses normal histograms. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#56 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AGyTdtDCazDxVCybfymiiM1uEi9b9_BBks5uP0yRgaJpZM4P1tPe> .