223692 - Don't use the initial-response argument with AUTH LOGIN

Status: RESOLVED FIXED

(MailNews Core :: Networking: SMTP, defect)

Description

[Christian Eyrich]

I just discovered or better got confirmed that the SMTP client shouldn't use the
 initial-response argument with AUTH LOGIN.

"The original design of LOGIN by Mark Crispin (and what we documented) does NOT
permit the client to use an initial response." The documentation (currently as
draft) is available through the above URL.
Read RFC 2554 too: "The optional initial-response argument to the AUTH command
is used to save a round trip when using authentication mechanisms that are
defined to send no data in the initial challenge." But the LOGIN mechanism sends
data in the initial challenges.

Although much server implementations allow the initial-response argument we
shouldn't depend on it.

Comment 1

[Christian Eyrich]

Attachment: proposed patch

This patch changes Mozillas behaviour not to use the initial-response argument
with LOGIN mechanism.

I also commented out the (anyway not used) old LoginResponse(). Basically to
use the name for one of the new functions needed.

Comment 2

[David :Bienvenu]

+    if (m_responseCode/100 != 3)
+        m_nextState = SMTP_AUTH_LOGIN_RESPONSE;
+    else
+        m_nextState = SMTP_SEND_AUTH_LOGIN_USERNAME;
can be simply:

m_nextState = (m_responseCode/100 == 3) ? SMTP_SEND_AUTH_LOGIN_USERNAME :
SMTP_AUTH_LOGIN_RESPONSE;

can you remove the commented out code instead of just commenting it out? Other
than that, it looks OK.

Comment 3

[Christian Eyrich]

Attachment: patch v2

Replaced the if from v1 (seems I'm a sucker for this kind of if). I also
removed the commented out code. I let it in just in case.

Comment 4

[David :Bienvenu]

Comment on attachment 135130 [details] [diff] [review]
patch v2

looks good - I'll fix the tabs before I check in.

Comment 5

[David :Bienvenu]

fix checked in.