Fred Hebert

The [README for vegur](https://github.com/heroku/vegur/) contains all relevant information for vegur. [This library's own README](https://github.com/heroku/cowboyku) contains relevant information to the forking of cowboy (with my own emphasis added): > Cowboy is...

why not add `unpack_file` ? It would allow you to carry the intent ("the binary blob is actually a filename") in the function name itself, rather than asking the user...

For example, you know if the codes are in 5xx range that the error is not on the client's side; it is generally safe to retry any operation multiple times...

Yeah that's all fair. I was just saying that those are inherent things to the HTTP protocol (as defined in its spec)

No there's no good reason to remove whitespace from a path, it just makes it wrong. The proper fix is to escape and/or quote the path adequately.

You might at least want to try a newer Rebar3, we'Re up to 3.20.

The first error you get is likely to be bypassed if you use the master version of rebar3. We should cut a release soon.

it's trying to add a version to the end of the filepath and failing because it's a binary instead of a list. I can't recall if that was patched in...

right, because the case expression is fully local, the branch is unique to each call-site. I'm noticing that we already do some dynamic dispatch on `otel_counter:add/4`: https://github.com/open-telemetry/opentelemetry-erlang/blob/e6c5c756d9a5df3714c5f379fcdae3ab1fae7980/apps/opentelemetry_api_experimental/src/otel_counter.erl#L40-L48 I don't think...

An alternative is to make it a sort of "private API" function that the macro uses to create a layer of indirection but passes the union of all required context...