Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > comp.lang.java.programmer > #6668

Re: JavaMail bug?

From Martin Gregorie <martin@address-in-sig.invalid>
Newsgroups comp.lang.java.programmer
Subject Re: JavaMail bug?
Date 2011-07-29 21:32 +0000
Organization UK Free Software Network
Message-ID <j0v8t7$os9$1@localhost.localdomain> (permalink)
References <j0fb5e$sat$1@localhost.localdomain> <995djcFq0nU1@mid.individual.net> <j0khlj$8bg$3@localhost.localdomain> <j0kilj$8bg$4@localhost.localdomain>

Show all headers | View raw


On Mon, 25 Jul 2011 20:11:31 +0000, Martin Gregorie wrote:

> I meant to add that I'm sorting out a bit of C at present. Running your
> test program and extracting my SSCE it on the list behind it.
>
Sorted. Now back the the JavaMail problem:

I modified the TestMail SSCE to suit my environment by changing the 
'mail.smtp.host' property to "ukfsn.org" (my ISP) and the transport 
connect statement became

    tr.connect("zoogz.gregorie.org", "kiwi", "n/a")

This works regardless of whether the local MTA is running or stopped.

Then I made my own SSCE, MATestMail, from code in my mail dispatch class. 
The only differences between my SSCE and TestMail are that it does not 
add the 'mail.smtp.host' property to system props and it gets the session 
with "Session.getInstance(props)" rather than "Session.getDefaultInstance
(props, null)".

It also creates and sends a message to myself so the headers can be seen 
and I added a quick and dirty hack to optionally turn on SMTP debugging. 
Both turned out to be useful for working out exactly what was going on.

The 'gotcha' turned out to be my omission to set the 'mail.smtp.host' 
property. I didn't do that because the connect() documentation says that 
it overrides the default host. This turns out to be incorrect: the MTA 
host named in connect() appears in all the debugging and is used by the 
getURL method, *but* the hostname supplied in the 'mail.smtp.host' 
property is what defines the MTA that is actually used. If the property 
is omitted the MTA host name defaults to localhost.

When I added that property and messed about with its value I found the 
following behaviour, which was clearly shown by looking at the test 
message headers and the /var/log/maillog on zoogz:

- With the 'mail.smtp.host' property unset the local MTA is always used:
  if its running the message is sent via 'localhost' and if it is stopped
  JavaMail reports a connection failure.

- With it set to 'mail.ukfsn.org' the mail went via my ISP's mail server.

- With it set to 'zoogz.gregorie.org' the mail is sent directly to zoogz.

During all three property values the connect() method was using 
"zoogz.gregorie.org" as the 'host' parameter.

For completeness, here's my SSCE:

import java.util.Date;
import java.util.Properties;
import javax.mail.*;
import javax.mail.internet.*;

public class MATestMail
{
   public static void main(String args[])
   {
      String         host      = "zoogz.gregorie.org";
      String         user      = "kiwi";
      String         password  = "n/a";
      String         to        = "martin@gregorie.org";
      String         from      = "ma@gregorie.org";
      String         subject   = "Test message";
      boolean        debug     = true;
      Session        sesh;

      try
      {
         Properties props = System.getProperties();

         sesh = Session.getInstance(props);
         if (args.length > 0)
            sesh.setDebug(debug);
      
         MimeMessage    msg  = new MimeMessage(sesh);
         MimeMultipart  body = new MimeMultipart();
         InternetAddress[] recipients = {new InternetAddress(to)};
         InternetAddress   sender = new InternetAddress(from);
         msg.setFrom(sender);
         msg.setRecipients(Message.RecipientType.TO, recipients);
         msg.setSubject(subject);
         msg.setSentDate(new Date());
         MimeBodyPart b = new MimeBodyPart();
         b.setText("This is an SSCE test message\n");
         body.addBodyPart(b);
         msg.setContent(body);
         msg.saveChanges();

         Transport tr = sesh.getTransport("smtp");
         tr.connect(host, user, password);
         System.out.println(tr);
         System.out.println(tr.getURLName());
         tr.send(msg);
         tr.close();
      }
      catch(MessagingException e)
      {
         e.printStackTrace();
      }
   }
}

This SSCE doesn't set 'mail.smtp.host' and so will fail if the localhost 
MTA isn't running: I develop on a Linux box, so there's always a local 
MTA - hence the comment about the localhost MTA not running rather than 
it not existing.

To make this SSCE connect directly to the target MTA (zoogz) I added the 
statement:

	props.put("mail.smtp.host", "zoogz.gregorie.org");

immediately after the 'Properties props = System.getProperties();' 
statement.

The description of Transport.connect(host, user, password) is not at all 
clearly written, but having read it and the description of 
Transport.connect(user, password) I still think it means that the values 
supplied as connect() parameters should take precedence over those set 
via the property system, however connect() is plainly written to do the 
opposite, leaving me to wonder why connect(host, user, password) even 
exists.

So, what do you think: have I found a bug?

Many thanks to Nigel and Knute for helpful suggestions and example code.

-- 
martin@   | Martin Gregorie
gregorie. | Essex, UK
org       |

Back to comp.lang.java.programmer | Previous | NextPrevious in thread | Next in thread | Find similar


Thread

JavaMail bug? Martin Gregorie <martin@address-in-sig.invalid> - 2011-07-23 20:32 +0000
  Re: JavaMail bug? Knute Johnson <september@knutejohnson.com> - 2011-07-23 20:54 -0700
    Re: JavaMail bug? Martin Gregorie <martin@address-in-sig.invalid> - 2011-07-24 09:50 +0000
      Re: JavaMail bug? Knute Johnson <september@knutejohnson.com> - 2011-07-24 09:27 -0700
        Re: JavaMail bug? Martin Gregorie <martin@address-in-sig.invalid> - 2011-07-25 19:36 +0000
          Re: JavaMail bug? Knute Johnson <nospam@rabbitbrush.frazmtn.com> - 2011-07-25 23:00 -0700
            Re: JavaMail bug? Steve Sobol <sjsobol@JustThe.net> - 2011-07-25 23:18 -0700
              Re: JavaMail bug? Martin Gregorie <martin@address-in-sig.invalid> - 2011-07-26 17:49 +0000
            Re: JavaMail bug? Martin Gregorie <martin@address-in-sig.invalid> - 2011-07-26 17:42 +0000
  Re: JavaMail bug? Nigel Wade <nmw-news@ion.le.ac.uk> - 2011-07-25 15:40 +0100
    Re: JavaMail bug? Martin Gregorie <martin@address-in-sig.invalid> - 2011-07-25 19:54 +0000
      Re: JavaMail bug? Martin Gregorie <martin@address-in-sig.invalid> - 2011-07-25 20:11 +0000
        Re: JavaMail bug? Martin Gregorie <martin@address-in-sig.invalid> - 2011-07-29 21:32 +0000
          Re: JavaMail bug? Knute Johnson <september@knutejohnson.com> - 2011-07-29 16:26 -0700
          Re: JavaMail bug? Nigel Wade <nmw-news@ion.le.ac.uk> - 2011-08-01 13:20 +0100
            Re: JavaMail bug? Martin Gregorie <martin@address-in-sig.invalid> - 2011-08-01 21:49 +0000
              Re: JavaMail bug? Nigel Wade <nmw-news@ion.le.ac.uk> - 2011-08-02 16:34 +0100
                Re: JavaMail bug? Martin Gregorie <martin@address-in-sig.invalid> - 2011-08-02 20:01 +0000
                Re: JavaMail bug? Nigel Wade <nmw-news@ion.le.ac.uk> - 2011-08-03 09:27 +0100
                Re: JavaMail bug? Martin Gregorie <martin@address-in-sig.invalid> - 2011-08-08 22:50 +0000
                Re: JavaMail bug? Nigel Wade <nmw-news@ion.le.ac.uk> - 2011-08-09 13:31 +0100
                Re: JavaMail bug? Martin Gregorie <martin@address-in-sig.invalid> - 2011-08-09 22:47 +0000
                Re: JavaMail bug? Nigel Wade <nmw-news@ion.le.ac.uk> - 2011-08-10 09:34 +0100
                Re: JavaMail bug? Martin Gregorie <martin@address-in-sig.invalid> - 2011-08-10 19:37 +0000
                Re: JavaMail bug? Nigel Wade <nmw-news@ion.le.ac.uk> - 2011-08-03 09:38 +0100
                Re: JavaMail bug? Martin Gregorie <martin@address-in-sig.invalid> - 2011-08-16 21:06 +0000
                Re: JavaMail bug? Nigel Wade <nmw-news@ion.le.ac.uk> - 2011-08-17 10:29 +0100
                Re: JavaMail bug? Martin Gregorie <martin@address-in-sig.invalid> - 2011-08-17 19:54 +0000
          Re: JavaMail bug? Andreas Leitgeb <avl@gamma.logic.tuwien.ac.at> - 2011-08-01 12:54 +0000

csiph-web