I still disagree with that stance and I've reached out to clarify. Leaving this here until we find an agreement

@iMacTia I can try to help here. Is there any simple reproducer for faraday-net_http case with Ruby 2.6? I tried locally and everything works fine in my simple scripts 🤔.

I think this only happens where there's another gem directly pulling in a standard gem as a direct dependency.
You could try using your branch here (with net-http as a Faraday dependency and requiring net-http or net-protocol before faraday.

# on simi:patch-1 branch
require 'net-http' # (or net-protocol?)
require 'faraday-net_http'

I'm not sure though, I haven't experienced this issue directly myself yet

If that doesn't work, then mixing faraday-net_http with mail_room seems to cause the issue

OK, thanks for the hints. I'll try to reproduce and report back.

Just to clarify, I'm 👍 to this PR and I wouldn't expect it to cause issues on any supported rubies if the minimum supported version is 2.6, and net/http was gamified in Ruby 2.5.

Since Ruby 2.6 is EOL now, would it make sense to bump required ruby version and finally fix this?

We've recently made Ruby 3.0 the minimum for Faraday (change is in main, to be released).

Since net-http became a default gem in 3.0, I'm now happy to get this merged and released after bumping the minimum ruby version to 3.0. Would that work for you @simi ?

Let's ensure with @deivid-rodriguez this is still good idea. From my side 👍.

I'll get things started while we wait for @deivid-rodriguez (merge + CI changes), will hold on cutting a release though 🙂

Yes, I still think this is a good idea, and also smoother if the version that ships the change won't be installable on older rubies 👍.

@deivid-rodriguez @simi thank you both for participating in this discussion and for your patience.
faraday-net_http 3.1 has now been released (thanks to @olleolleolle) as well as Faraday 2.9.0 which allows to use this new release.

If you could give it a try, your feedback would be greatly appreciated 🙏

@deivid-rodriguez I know our environment is a bit weird, but I think this simple dependency addition exposed a possible RubyGems bug: ruby/rubygems#7374