I did this

The application does a regular GET of an HTTP URL:

  test_setopt(curl, CURLOPT_HTTPAUTH, CURLAUTH_ANY);
  test_setopt(curl, CURLOPT_USERNAME, "user");
  test_setopt(curl, CURLOPT_PASSWORD, "password");
  test_setopt(curl, CURLOPT_REDIR_PROTOCOLS_STR, "imap");

The server responds with a 401 and regular WWW-Authenticate: response headers.

libcurl fails to get the URL.

I expected the following

Successful retrieval of the page. Removing the CURLOPT_REDIR_PROTOCOLS_STR line makes it succeed again.

curl/libcurl version

current git master, but this is a bug that probably has always existed.

operating system

Linux, but is platform agnostic.