Aliaksandr Valialkin

@AndrewChubatiuk , it is better from usability perspective to handle all the HTTP requests starting from `insert/elasticsearch/_template/` in a generic way at VictoriaLogs side, so client won't notice these paths...

This issue should be fixed in [v1.102.0-rc1 release](https://github.com/VictoriaMetrics/VictoriaMetrics/releases/tag/v1.102.0-rc1).

FYI, see the follow-up commit 4f99799db706790af7fd79a47d0c00ae720af006

The issue should be fixed in [v1.97.5 LTS release](https://github.com/VictoriaMetrics/VictoriaMetrics/releases/tag/v1.97.5) and in [v1.102.0-rc1 release](https://github.com/VictoriaMetrics/VictoriaMetrics/releases/tag/v1.102.0-rc1). @ebensom , please re-open the issue if it still persists in these releases of [vmauth](https://docs.victoriametrics.com/vmauth.html).

Iniflags doesn't support multi-line values at the moment. But it may be implemented in the future quoting such values in backticks as shown in your example.

Agreed with the reasoning. Unfortunately, the current behaviour cannot be changed in existing software like VictoriaMetrics, since this will break expectations of many users, who rely on the current behaviour...

> If it supports the Influx line protocol, then it should be supported by k6 already, no? Yes - it should work OK on data ingestion side. As for data...

It would be great analyzing top production use cases for weak pointers in other programming languages before deciding on how to implement them in Go. Probably, it would be better...

See https://github.com/VictoriaMetrics/VictoriaLogs/issues/323

[This article](https://valyala.medium.com/go-evolves-in-the-wrong-direction-7dfda8a1a620) may be useful during the investigation.