Max Inden
Max Inden
@cheshirskycode any updates here? Anything I can help with?
> As I want dynamic buckets Is this supported by Prometheus? As far as I know, Prometheus expects Histogram buckets to be constant throughout the lifetime of a monitoring target.
Good point @blckngm. Thanks for reporting. Yes. I wonder whether we should actively escape them or simply panic (in debug mode!) when they are not escaped. Preference for the latter....
Fine with escaping one character at a time as long as we have an accompanying benchmark that proves that this is not a performance bottleneck.
Very much in favor of this. What do you think of adding this to the new `ConstCounter` and `ConstGauge` for now as neither of them would require synchronization and likely...
Why is `chrono` needed? Given that we need unix timestamps only, wouldn't `Instant` and `UNIX_EPOCH` be enough?
> One important note here: Prometheus requires millisecond timestamps while the OpenMetrics spec uses seconds. Since this is a Prometheus client not an OpenMetrics client it seems to be sensible...
Thank you for digging into this. Let's discuss on #175.
Thanks for the pull request. In order for me to better understand the feature request, mind expanding on your concrete use-cases for this method?
Closing here with the concerns above. Feel free to open an issue in case you want to continue the discussion.