59634 - remove the try_ssl code, it's auth_method only now.

Status: VERIFIED INVALID

(MailNews Core :: Networking: SMTP, defect, P3)

Description

[Jarrod Gray]

I have my smtp try_ssl pref set to 2, which should make smtp always use ssl to 
send mail. I sniffed my packets and there was no ssl transaction - everything 
went through plain text. I did the same thing with NS4.75 and the packets showed 
the ssl transaction.

I'm running
Mozilla/5.0 (Windows; U; Win98; en-US; m18) Gecko/20001109
a trunk build.

Comment 1

[jefft]

What pref name are you using? The pref has been changed. Try the following and
let me know if this fix your problem.

user_pref("mail.smtpserver.smtp1.auth_method", 3);

0 -- no auth
1 -- auth any
2 -- auth login
3 -- try ssl/tls
4 -- ssl/tls only

Comment 2

[Jarrod Gray]

I am using try_ssl because of what I've seen here in nsSmtpProtocol.h. It looks
to me that by using auth_method I would be requesting to use TLS and not SSL.

 89 typedef enum _PrefAuthMethod {
 90 PREF_AUTH_NONE = 0,
 91 PREF_AUTH_ANY = 1,
 92 PREF_AUTH_LOGIN = 2,
 93 PREF_AUTH_TLS_TRY = 3,
 94 PREF_AUTH_TLS_ONLY = 4
 95 } PrefAuthMethod;
 96
 97 typedef enum _PrefTrySSL {
 98 PREF_NO_SSL = 0,
 99 PREF_TRY_SSL = 1,
100 PREF_ALWAYS_SSL = 2
101 } PrefTrySSL;

Comment 3

[jefft]

PrefTrySSL is 4.7 enum which should go away. TLS is a synonum to SSL.

Comment 4

[(not reading, please use [e-mail] instead)]

I've told jeff that I'll take care of removing the try_ssl stuff.

updating summary, taking bug from jeff.

the only concern I have before I do that is migration.  I'm working on that
issue and I'll have it resolved soon.

Comment 5

[Azrael AOD]

Is the reporter's case valid in that this is a bug? Hence is it confirmed, or 
invalid?
Seth: seeing how you took the bug for yourself.. does that mean you can confirm 
it?

Comment 6

[Keyser Sose]

Marking INVALID due to lack of response. Reopen if its still a problem in the
latest nightlies.

Comment 7

[(not reading, please use [e-mail] instead)]

reopening.

this issue still needs to be investigated.

Comment 8

[Duncan Findlay]

well, seth, would you like to confirm it?

Comment 9

[(not reading, please use [e-mail] instead)]

confirming.

putterman, who'd we give the SMTP / SSL ui issue to?

they should get this one, too.

Comment 10

[scottputterman]

what's the actual bug here? Is it not working even with jefft's suggestion, or
is this just to remove an unused enumeration?

Comment 11

[(not reading, please use [e-mail] instead)]

I'm not sure what the deal is anymore.

at one point, jefft told me that the per server pref of "try_ssl" should go
away, we were going to use the per server pref of "auth_method".

if this is the case, there is C++ code and some UI fragments smtpEditOverlay.*
that need to be removed and/or cleaned up.

on a related note, I'm not sure if the per server "auth_method" pref is the way
to go.  are those 5 values for auth_method the ones we want?  do they match up
to what the back end can (will or should be able to) do?  do they cover all the
cases we want to cover?

once we figure that out, we can design and implement a ui to allow users to
configure their SMTP servers properly.

my suggestion is to assign all the SMTP SSL / Auth, SMTP UI bugs to one person,
and let them research and sort it out.

re-assigning to putterman.

Comment 12

[John G. Myers]

The SSL values in auth_method are gone.  Authentication and SSL are orthogonal 
issues.

Comment 13

[Jarrod Gray]

When sending email I always get an error "Send of message failed" with my auth
set to ssl always. When I sniff the transaction, I see the server send the "go
ahead with TLS transaction" and the client begins but the message still fails.
Are you seeing this behavior jgmyers?

Comment 14

[esther]

The last comment (and problem) in this bug is a dup of 65703, and since the
preference mentioned in the original scenario is invalid, verifying as invalid.