David Ashpole

I looked at the implementation, and it looks like it should have been sending the histogram-portion of the metric correctly already. The error described above just drops the summary point.

I opened https://github.com/open-telemetry/opentelemetry-collector-contrib/pull/13317 to improve the error message, but I don't think there is more we can do

The description of `/stacktrace` in the docs you reference looks like it only applies to the v1 API, while we use the v2 API: `JSON-formatted stack trace. Stack traces aren't...

I don't know if that would be a great contract for the exporter to have. It would be fairly difficult for clients to implement correctly, so I would prefer having...

A teammate pointed me to https://cloud.google.com/trace/docs/quotas, which says we are limited to 32 attributes per-span. I'm worried that the multiple-attribute approach would hit that limit ~instantly, and just end up...

I could totally see that working for an SDK exporter (e.g. provide a string -> StackTrace function), but that sounds hard to do in the collector via yaml config.

It seems really hard for users to implement... I'll have to chew on it a bit. Just for fun, can you try adding this in your exporter config: ```yaml trace:...

Not right now. If we did go that route it is likely something we would have to keep around (we wouldn't risk breaking users by deprecating it), so i'd like...

cc @open-telemetry/wg-prometheus

> Is there a test that checks the new scenario? I've restored the existing unit test without an inf bucket.