incorrect behavior for the Sender field

incorrect behavior for the Sender field

[Hunter Kelly]

I know that there has been a lot of discussion about whether a user
should be allowed to change the "Sender" field, but that is irrelevant
to the problem that I am having.

Red Gnus 0.74 (and earlier versions -- I am not sure how far back this
has been happening) generates a sender field when it shouldn't be.

The mail that I am sending out has headers that look like this:

Return-Path: retnuh@pixar.com
Received: from corona.pixar.com (corona.pixar.com [138.72.20.84]) by scam.XCF.Berkeley.EDU (8.7.5/8.7.3) with SMTP id SAA18597 for <retnuh@xcf.berkeley.edu>; Mon, 9 Dec 1996 18:01:46 -0800
Received: by corona.pixar.com (Smail3.1.29.1 #2)
        id m0vXHa6-00011dC; Mon, 9 Dec 96 18:06 PST
Sender: retnuh@pixar.com (Hunter Kelly)
Sender: retnuh@corona
To: retnuh@scam.XCF.Berkeley.EDU
Subject: test 7 methinks
From: Hunter Kelly <retnuh@corona.pixar.com>
Date: 09 Dec 1996 18:03:15 -0800
Message-ID: <yvt3exf40m4.fsf@corona.pixar.com>
Lines: 3


Notice that there are two lines that are "Sender".  I am not inserting
my own headers in any way.  The two occurances are causing messages to
get trashed in other mail readers, like Netscape 2.0.

I have no idea why it is generating the Sender field twice.  But the
second field is wrong, anyway.  Here are some relevant variables.

(message-make-sender)
"retnuh@corona"

(message-make-address)
"retnuh@corona.pixar.com"

(message-user-mail-address)
"retnuh@corona.pixar.com"

mail-host-address
"corona.pixar.com"


Now, in message-generate-headers, there is a block of code that looks
like this:

	(when (and from 
		   (not (message-check-element 'sender))
		   (not (string=
			 (downcase
			  (cadr (mail-extract-address-components from)))
			 (downcase secure-sender)))
		   (or (null sender)
		       (not 
			(string=
			 (downcase
			  (cadr (mail-extract-address-components sender)))
			 (downcase secure-sender)))))

I believe that the second string= is the wrong test.  It should be some form
of string match, because, as it has shown in my case, I am not
fiddling with the headers, but system-name does not return the FQDN.

This is really, really annoying, because certain people cannot read
mail I send them.  I think that either this needs to get fixed, so the
above behaves correctly, or Gnus should stop second-guessing the user
by insisting on a real Sender field.


Hunter

Re: incorrect behavior for the Sender field

[Rich Pieri]

-----BEGIN PGP SIGNED MESSAGE-----

>>>>> "HK" == Hunter Kelly <retnuh@corona.pixar.com> writes:

HK> Sender: retnuh@pixar.com (Hunter Kelly)
HK> Sender: retnuh@corona

Your MTA (the one on corona) is inserting the "first" header, which it
should not do.  First, when "overwriting" an existing header, the
existing header should be removed.  Second, Sender is supposed to
somehow point back at the mailbox on the machine responsible for the
initial submission to the network; in this case it is corona, or
corona.pixar.com, unless pixar.com *is* corona.pixar.com (which it is
not).

So, I would look into getting your MTA fixed.  Since it is doing
something broken, there is a good chance that it is doing quite a bit
more that is broken.

-----BEGIN PGP SIGNATURE-----
Version: 2.6.2
Charset: noconv

iQCVAwUBMq19/J6VRH7BJMxHAQFdrAP/UdaB57ij+2E7cNWlySSk1zHFfLFl78I4
WHl2/0beOcGfQrJEkDSTCn7KX0SaAZrkwlcxo1oDuShnxUbaYcg/9d9eV7cn4qzq
i8qh+GhDQK5idKzRqtwqE2uqRZXYk6E9M3+v8SXvBB5jaL25UMAbl0sYqbL7lhPK
TPYFIClI4Ho=
=gumN
-----END PGP SIGNATURE-----
-- 
Rich Pieri <rich.pieri@prescienttech.com> | When eating out, think nothing of
      Prescient Technologies, Inc.        | sending back your meal twenty or
       A Stone & Webster Company          | thirty times.
  I speak for myself, not PTI or SWEC     | --A cat's guide to life

Re: incorrect behavior for the Sender field

[Hunter Kelly]

Rich Pieri <rich.pieri@PrescientTech.com> writes:

> -----BEGIN PGP SIGNED MESSAGE-----
> 
> >>>>> "HK" == Hunter Kelly <retnuh@corona.pixar.com> writes:
> 
> HK> Sender: retnuh@pixar.com (Hunter Kelly)
> HK> Sender: retnuh@corona
> 
> Your MTA (the one on corona) is inserting the "first" header, which it
> should not do.  First, when "overwriting" an existing header, the
> existing header should be removed.  Second, Sender is supposed to
> somehow point back at the mailbox on the machine responsible for the
> initial submission to the network; in this case it is corona, or
> corona.pixar.com, unless pixar.com *is* corona.pixar.com (which it is
> not).
> 
> So, I would look into getting your MTA fixed.  Since it is doing
> something broken, there is a good chance that it is doing quite a bit
> more that is broken.

You missed my point entirely.  Gnus should insert a Sender field that
is correct, ie retnuh@corona.pixar.com.  Inserting a field
retnuh@corona is totally useless.  And it shouldn't be inserting a
"Sender" field at all, because "retnuh@corona" is a totally valid
substring of "retnuh@corona.pixar.com", which is what the "From" field
states.

Hunter

Re: incorrect behavior for the Sender field

[Rich Pieri]

-----BEGIN PGP SIGNED MESSAGE-----

>>>>> "HK" == Hunter Kelly <retnuh@corona.pixar.com> writes:

HK> You missed my point entirely.  Gnus should insert a Sender field
HK> that is correct, ie retnuh@corona.pixar.com.

"retnuh@corona.pixar.com" may not be correct.  Actually, it probably is
correct, but system-name is not returning the correct value.  This is
probably because corona is not using its full host name as its host
name.  That is, corona knows itself as "corona" rather than
"corona.pixar.com" -- and I would bet that corona is running some flavor
of SunOS because of this.  The way to fix that is to change corona's
host name (and as a side effect this will fix any problems you have with
local mail having the domain part stripped).

HK> Inserting a field retnuh@corona is totally useless.

Not totally useless, but at least mostly useless.  But this is not Gnus'
fault; do not blame it for system misconfiguration.  If your system is
not configured to return a full host name when asked for one, it is the
system that is misbehaving, not the software making the request.

HK> And it shouldn't be inserting a "Sender" field at all, because
HK> "retnuh@corona" is a totally valid substring of
HK> "retnuh@corona.pixar.com", which is what the "From" field states.

Nope.  Sender gets inserted when an exact match fails.  If what the MUA
decides is the "real" name of the host is at all different from what is
in the From header, a Sender header gets generated.  Gnus is at least a
little smart about it by comparing just the mailbox portion of the From
header with `gnus-user-name'@`system-name', which is all that really
needs to be compared.

-----BEGIN PGP SIGNATURE-----
Version: 2.6.2
Charset: noconv

iQCVAwUBMq27Yp6VRH7BJMxHAQEywgP/WPElNKoxb1rgc5WLppxHE1e65BRv/pUe
fQI+r4nAGWJVsE40WrGaCi/WTyHzn5OXJZmjS6N+wPSYHph0NbrWAtvXYXkwZf23
6acy820+kD5+oqkjJKtD3Jb1JQyFxqTFTh+taxnOVMeGjWdK/19f6gXFh2jwO7Dl
2pvCWujU4PU=
=mWhz
-----END PGP SIGNATURE-----
-- 
Rich Pieri <rich.pieri@prescienttech.com> | When eating out, think nothing of
      Prescient Technologies, Inc.        | sending back your meal twenty or
       A Stone & Webster Company          | thirty times.
  I speak for myself, not PTI or SWEC     | --A cat's guide to life