On Thu, 28 May 2020 at 19:52, Bjoern Rabenstein <bjo...@rabenste.in> wrote:
> Dear Prometheans, > > So far, we have recommended Celsius as the base unit for temperatures, > despite Kelvin being the SI unit. That was well justified by the > overwhelming majority of use cases, where Kelvin would be just > weird. I'd really like to see more scientific usage of Prometheus, so > I was never super happy with that recommendation, but since it was > just a recommendation, I could live with it. > > Now Matt Layher came up with another, more technical use case: color > temperature. Here, using Celsius would be even weirder. So there is a > case where you clearly do not want to follow the suggestion of the > linter, which is more in line with typical Prometheus use cases than > my arguably somewhat far fetched time series for low-temperature > experiments. > > Therefore, Matt suggested to make the metrics linter not complain > about kelvin. > > I think this is a clearly defined problem with clear arguments and a > clear disagreement between Brian Brazil on the one side and Matt and > myself on the other side. The appropriate amount of effort has been > spent to find a consensus. I disagree that sufficient effort has been made to find concensus, and in particular I don't feel that my objection that this change would have wider ranging consequences for the best practices of metrics names as a whole has been addressed. Jumping so quickly to a vote also makes it difficult to discover if maybe a concensus could have been found here. The question has not been asked that if there are valid use cases where an exception makes sense, does that now mean that the linter must never complain about that unit as whole? In case you aren't aware the node exporter exposes the size of the entropy pool in bits - which I'm okay with along roughly the lines you're arguing on - but yet the linter does not accept bits. The linter is a recommendation, not a hard rule that must be followed. One potential concensus might be around adding the nuance/clarification in the exporter guidelines where this is covered (it already kinda says it, but it takes some squinting). My vote is NO, as I disagree with the linter accepting any more than one base unit. Brian All arguments can be found in > https://github.com/prometheus/client_golang/pull/761 and > https://github.com/prometheus/docs/pull/1648 . > > I hereby call a vote for the following proposal: > > Allow Kelvin as a base unit in certain cases and update our > documented recommendation and the linter code accordingly. > > > (The changes may take the form of the two PRs out there, but the vote > in about the general idea above, not the implementation detail.) > > > The vote closes on 2020-06-04 20:00 UTC. > -- > Björn Rabenstein > [PGP-ID] 0x851C3DA17D748D03 > [email] bjo...@rabenste.in > > -- > You received this message because you are subscribed to the Google Groups > "Prometheus Developers" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to prometheus-developers+unsubscr...@googlegroups.com. > To view this discussion on the web visit > https://groups.google.com/d/msgid/prometheus-developers/20200528185216.GK2326%40jahnn > . > -- Brian Brazil www.robustperception.io -- You received this message because you are subscribed to the Google Groups "Prometheus Developers" group. To unsubscribe from this group and stop receiving emails from it, send an email to prometheus-developers+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-developers/CAHJKeLopqQ%3Dc_dhjSFG8kC6G4-82nxmEmN1PJGPfs8qRyjjxpw%40mail.gmail.com.
