Hey James,

Funny, I stumbled upon this as I was opening my own PR for this exact MAAS api issue!

I opened #1597 explaining the behavior a bit but wanted to comment on this PR rather than opening a duplicate:

• I recommend leaving the OAuth consumerSecret parameter as a regular string vs the nullable type you changed it to. This should reduce the impact of this change as it would be a signature change vs an internal validation change.

• It might also be beneficial to modify the unit test to account for the different OAuthTypes:

[Test]
[TestCase(OAuthType.AccessToken)]
[TestCase(OAuthType.ProtectedResource)]
[TestCase(OAuthType.AccessToken)]
[TestCase(OAuthType.ProtectedResource)]
public void Authenticate_ShouldAllowEmptyConsumerSecret_OnHttpAuthorizationHeaderHandling(OAuthType type)
{
    // Arrange
    const string url = "https://no-query.string";

    var client = new RestClient(url);
    var request = new RestRequest();
    _authenticator.Type           = type;
    _authenticator.ConsumerSecret = "";

    // Act
    _authenticator.Authenticate(client, request);

    // Assert
    var authParameter = request.Parameters.Single(x => x.Name == "Authorization");
    var value = (string)authParameter.Value;

    Assert.IsNotEmpty(value);
    Assert.IsTrue(value!.Contains("OAuth"));
    Assert.IsTrue(value.Contains("oauth_signature=\"" + OAuthTools.UrlEncodeStrict("&")));
}

Hello Patrick,

Many thanks for your feedback and I am happy to see some tractions around this use case! I have applied your suggestion regarding the enhancement of the unit test.

I recommend leaving the OAuth consumerSecret parameter as a regular string vs the nullable type you changed it to. This should reduce the impact of this change as it would be a signature change vs an internal validation change.

I hesitated before changing these signatures. My reasoning was that introducing a nullable type for the parameter consumerSecret would be ok, since C# allowed an implicit cast from string? to string and this conveys the intent that the consumer secret may be empty. In the case I restore the initial type of the parameter (string), would you still allow passing null as an argument to consumerSecret, or only allow an empty string to be passed? Should we wait the feedback of a RestSharp's dev before changing it?

Cheers!