Björn Rabenstein

In the meantime, paging @roidelapluie again: It would be very good to get a general call about the changes in the API as discussed above and previously on [Slack](https://cloud-native.slack.com/archives/C01AUBA4PFE/p1692127732941169). Currently,...

(Disclaimer: I still have to do the detailed review.) I hijacked an unrelated meeting with @roidelapluie to find out what he thinks about this. He brought up the concern that...

We already have https://github.com/prometheus/prometheus/issues/12945 and https://github.com/prometheus/prometheus/pull/12922 . Please discuss there. (I'm pretty sure we should not warn on nameless metrics. Once we have stronger typing in PromQL, we can do...

@ashutosh887 Awesome. The first step would be, as said, to do some research and find out all the features of `promtool` that need work to support native histograms. A bit...

When it comes to the actual implementation of the items on the (still to be created) checklist, it should be noted that the rules unit tests use the same testing...

The rules unit tests now support native histograms. :tada:

Thanks for the great analysis. Note that there is also #12331 (which is more about creating a whole new sub-command and not really about adding support for native histograms to...

Yeah, sounds good. This change should also apply to conventional float chunks because their max chunk number can be configured by now. See #12055 .

https://github.com/prometheus/prometheus/pull/12668 should have added that capability. So my expectation is that `promtool` will run unit tests involving native histograms just fine. If not, that would be a bug.

Any ideas why it uses more RAM? With fewer chunks and a better compression ratio, I would expect _less_ RAM usage. If at all, I would expect more CPU usage...