When linux use https_proxy instead of http_proxy?

Question

I realized that in linux as well as in many programs, one can set http_proxy and https_proxy. I see many guides suggest:

http_proxy = http://host:port
https_proxy = https://host:port

I have serveral questions:

1. How a program determine which proxy, http or https, to connect? I tested in a docker alphine image to find out that wget get https://google.com through http_proxxy. It seems that this doesn't depend on which protocol to use.

2. What is the meaning of http, https in front of host:port? Does it suggest which protocol to use to connect to the proxy server? As I find out I can actually use http://host:port for https_proxy. Besides, setting without http

   http_proxxy = host:port

   also works.

Thank you!

Answer 1

It's pretty confusing I agree. Some programs use just http_proxy and adjust the protocol as needed for HTTP/HTTPS requests.

This is what https://www.npmjs.com/package/proxy-agent does anyway.

Some other programs use both. HTTPS_PROXY for https requests and HTTP_PROXY for http requests.

But I think it makes sense for most programs to just use a single HTTP_PROXY and adjust the protocol as needed. I can't see a good use case for using both and having separate proxies for HTTP vs HTTPS in this day and age.