cURL / Mailing Lists / curl-library / Single Mail


Re: Auth regression

From: Daniel Stenberg <>
Date: Tue, 23 Oct 2012 22:43:54 +0200 (CEST)

On Tue, 23 Oct 2012, Joe Mason wrote:

> I've added a new auth test at
> (top commit)
> It uses one connection and sends 6 requests on it, each with CURLAUTH_ALL.
> Each even-numbered request uses credentials "testuser:wrongpass" (will fail
> auth) and each odd-numbered request uses credentials "testuser:testpass".
> The server side of the test will reply to the first two requests with
> WWW-Authenticate: Basic (only), the next two with Digest (only), and the
> last two with NTLM (only). So, this tests that different types of auth can
> be used over the same connection when the auth type allowed by the user is
> not artificially restricted as in the previous auth tests.

I don't understand the test case. It seems to use a '1100' suffix, but there's
no <data1100> in the test2033 file. What is it supposed to do for 20331100 ?

List admin:
Received on 2012-10-23