Reply wasn't posted; will test new post

ruby-talk
21

"Thomas Fini Hansen" <beast@system-tnt.dk> schrieb im Newsbeitrag
> Oh, a world without Outlook.. As much as I'd love to see that, it's
> not an option either. Unless someone could come up with a gateway that
> fixes posts from Outlook. Secondly, there's webmails out there that
> has similar problems.

You're bashing the wrong pig here.

Perhaps, but I'm not particularly convinced.

"Outlook" itself does no news (at least older versions, I'm not
fully up to date). Instead it delegated news handling to "Outlook
Express" - a whole differnt piece of software (which I am using
btw).

Sounds reasonble to me, I'm not entirely up on Outlook either.

"Outlook Express" does a nice job inserting "References" headers,
which you can verify with this posting.

Yes, that's OE as a newsreader, but all my ranting was in the context
of O(E) as a mail client. Actually I figure it does a decent job as a
newsreader, but that'a not how most people use it.

But I did a quick investigation, in a post by Austin Ziegler:
X-MimeOLE: Produced By Microsoft Exchange V6.5.6944.0
And it has no In-Reply-To or References.. But the Tread headers. OK,
that's general MS evilness, there's no reason that because Outlook is
attached to an Exchange server, that it can't make proper mails when
sending to external addresses.

Then theres daz
X-Newsreader: Microsoft Outlook Express 6.00.2800.1106
And it looks good, but again, he's using it as a mailreader.

Digging through some private mail, I see some more of the Exchange
problem, and some Outlook/OE mail that do have the In-Reply-To, but no
References.

Secondly, I'm suspecting that O/OE is both using the same component to
create the mails:
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165

MS does a lot of evil, but they are not always the ones to blame.

No, but if the problem *is* related to the References header, then the
fact that Outlook forgets it in regular email, is an issue.

···

On Fri, Jun 04, 2004 at 06:18:43PM +0900, Robert Klemme wrote:

--
Thomas
beast@system-tnt.dk

22

fwiw: +1

···

il Thu, 3 Jun 2004 17:14:30 +0200, "Robert Klemme" <bob.news@gmx.net> ha scritto::

I could live with thread inconsistencies. At least we have all messages
of a thread available - even if not properly sorted. As said, that's my
favorite solution.

23

Saturday, June 5, 2004, 7:43:03 AM, Thomas Fini Hansen wrote:

"Outlook" itself does no news (at least older versions, I'm not
fully up to date). Instead it delegated news handling to "Outlook
Express" - a whole differnt piece of software (which I am using
btw).

Sounds reasonble to me, I'm not entirely up on Outlook either.

There are plugins for Usenet/NNTP handling for Outlook, but it is not
default.

But I did a quick investigation, in a post by Austin Ziegler:
X-MimeOLE: Produced By Microsoft Exchange V6.5.6944.0
And it has no In-Reply-To or References.. But the Tread headers. OK,
that's general MS evilness, there's no reason that because Outlook is
attached to an Exchange server, that it can't make proper mails when
sending to external addresses.

The problem could be the Exchange server, as its the one that has to
construct the outgoing Internet message from the Outlook format
object.

The problem is that if it is the Exchange server, then it does NOT
have the appropriate information to construct the In-Reply-To header
(References is a News header ONLY).

-austin

···

On Fri, Jun 04, 2004 at 06:18:43PM +0900, Robert Klemme wrote:

--
Best regards,
Austin mailto:austin@halostatue.ca

24

"Thomas Fini Hansen" <beast@system-tnt.dk> schrieb im Newsbeitrag
news:20040605114256.GA831@saber.xen.dk...

> "Thomas Fini Hansen" <beast@system-tnt.dk> schrieb im Newsbeitrag
> > Oh, a world without Outlook.. As much as I'd love to see that, it's
> > not an option either. Unless someone could come up with a gateway that
> > fixes posts from Outlook. Secondly, there's webmails out there that
> > has similar problems.
>
> You're bashing the wrong pig here.

Perhaps, but I'm not particularly convinced.

> "Outlook" itself does no news (at least older versions, I'm not
> fully up to date). Instead it delegated news handling to "Outlook
> Express" - a whole differnt piece of software (which I am using
> btw).

Sounds reasonble to me, I'm not entirely up on Outlook either.

> "Outlook Express" does a nice job inserting "References" headers,
> which you can verify with this posting.

Yes, that's OE as a newsreader, but all my ranting was in the context
of O(E) as a mail client. Actually I figure it does a decent job as a
newsreader, but that'a not how most people use it.

Ah, ok, I missed that one. Thought you were talking about news.

But I did a quick investigation, in a post by Austin Ziegler:
X-MimeOLE: Produced By Microsoft Exchange V6.5.6944.0
And it has no In-Reply-To or References.. But the Tread headers. OK,
that's general MS evilness, there's no reason that because Outlook is
attached to an Exchange server, that it can't make proper mails when
sending to external addresses.

Then theres daz
X-Newsreader: Microsoft Outlook Express 6.00.2800.1106
And it looks good, but again, he's using it as a mailreader.

Sounds like OE is the more mature product for the big wide world of internet
communication. AFAIK Outlook started out as an Exchange Server client, so,
yeah, I'd expect Outlook to be less compliant than OE.

Digging through some private mail, I see some more of the Exchange
problem, and some Outlook/OE mail that do have the In-Reply-To, but no
References.

Secondly, I'm suspecting that O/OE is both using the same component to
create the mails:
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1165

Depends on which part of the message is done by this component. I'd guess
it's rather mime formatting only but no headers.

> MS does a lot of evil, but they are not always the ones to blame.

No, but if the problem *is* related to the References header, then the
fact that Outlook forgets it in regular email, is an issue.

Indeed.

Thanks for clearing that up!

    robert

···

On Fri, Jun 04, 2004 at 06:18:43PM +0900, Robert Klemme wrote:

25

Saturday, June 5, 2004, 7:43:03 AM, Thomas Fini Hansen wrote:
> Sounds reasonble to me, I'm not entirely up on Outlook either.
There are plugins for Usenet/NNTP handling for Outlook, but it is not
default.

And I wont pass any judgement on that's compliance either, as I'm not
exposed to it.

> But I did a quick investigation, in a post by Austin Ziegler:
> X-MimeOLE: Produced By Microsoft Exchange V6.5.6944.0
> And it has no In-Reply-To or References.. But the Tread headers. OK,
> that's general MS evilness, there's no reason that because Outlook is
> attached to an Exchange server, that it can't make proper mails when
> sending to external addresses.

The problem could be the Exchange server, as its the one that has to
construct the outgoing Internet message from the Outlook format
object.

The problem is that if it is the Exchange server, then it does NOT
have the appropriate information to construct the In-Reply-To header

Yet, it's able to construct the Thread-* headers? It must know
*something*.

(References is a News header ONLY).

Wrong.

RFC 2822 - Internet Message Format:

# 3.6.4. Identification fields

···

On Sun, Jun 06, 2004 at 08:31:37AM +0900, Austin Ziegler wrote:
#
# Though optional, every message SHOULD have a "Message-ID:" field.
# Furthermore, reply messages SHOULD have "In-Reply-To:" and
# "References:" fields as appropriate, as described below.

--
Thomas
beast@system-tnt.dk

26

Thomas Fini Hansen wrote:

Austin Ziegler wrote:

> (References is a News header ONLY).

Wrong.

RFC 2822 - Internet Message Format:

RFC 2822 is a not a standard, yet.

From http://www.ietf.org/iesg/1rfc_index.txt (long)

"""
   2822 Internet Message Format. P. Resnick, Ed.. April 2001. (Format:
        TXT=110695 bytes) (Obsoletes RFC0822) (Status: PROPOSED STANDARD)
"""

A References: header is accepted by:
  RFC822: Standard for ARPA Internet Text Messages

and
  RFC850: Standard for Interchange of USENET Messages

  "... the rule is adopted that all USENET news articles
    must be formatted as valid ARPANET mail messages,
    according to the ARPANET standard RFC 822."

Our missing ML-> NG messages, after conversion from mail to news by rubygate,
are valid to RFC822/RFC850 standard regardless of the MUA used for posting.

daz

http://www.ruby-talk.org/102444

27

August 13, 1982

# 4.6.3. REFERENCES

···

On Mon, Jun 07, 2004 at 02:53:39AM +0900, daz wrote:

Thomas Fini Hansen wrote:
> Austin Ziegler wrote:
>
> > (References is a News header ONLY).
>
> Wrong.
>
> RFC 2822 - Internet Message Format:
>

RFC 2822 is a not a standard, yet.

>From http://www.ietf.org/iesg/1rfc_index.txt (long)

"""
   2822 Internet Message Format. P. Resnick, Ed.. April 2001. (Format:
        TXT=110695 bytes) (Obsoletes RFC0822) (Status: PROPOSED STANDARD)
"""

A References: header is accepted by:
  RFC822: Standard for ARPA Internet Text Messages

#
# The contents of this field identify other correspondence
# which this message references. Note that if message identif-
# iers are used, they must use the msg-id specification format.

The header is mentioned in a >20 year old spec, does contradict the
claim that it's a news header only, doesn't it? Yes, neither RFC says
that it is required, but the same goes for In-Reply-To. Both treat
them pretty much equal, so what anyone has to say about References,
goes for In-Reply-To as well.

and
  RFC850: Standard for Interchange of USENET Messages

  "... the rule is adopted that all USENET news articles
    must be formatted as valid ARPANET mail messages,
    according to the ARPANET standard RFC 822."

Our missing ML-> NG messages, after conversion from mail to news by rubygate,
are valid to RFC822/RFC850 standard regardless of the MUA used for posting.

That's reverse logic, it only claims that all news messages is valid
mails, not that all mail is valid news. And RFC850 in section 2.1.6
tells us that the References header is *required* when the Subject
header starts with 'Re: '. There's no such requirement for mail
messagees, or Outlook would be in *violation* of the specs, not just
being annoying.

But of course, all of this is just good practice, as no spec requires
either References nor In-Reply-To to be used.

--
Thomas
beast@system-tnt.dk

28

Thomas Fini Hansen wrote:

>
>
> A References: header is accepted by:
> RFC822: Standard for ARPA Internet Text Messages

August 13, 1982

The header is mentioned in a >20 year old spec, does contradict the
claim that it's a news header only, doesn't it?

I think so.

> and
> RFC850: Standard for Interchange of USENET Messages
>
> "... the rule is adopted that all USENET news articles
> must be formatted as valid ARPANET mail messages,
> according to the ARPANET standard RFC 822."
>
> Our missing ML-> NG messages, after conversion from mail to news by rubygate,
> are valid to RFC822/RFC850 standard regardless of the MUA used for posting.

That's reverse logic, it only claims that all news messages is valid
mails, not that all mail is valid news.

The claim that ruby-talk mail is converted by rubygate to RFC850 news
comes from me, supported by my (and others') empirical evidence.

daz

···

On Mon, Jun 07, 2004 at 02:53:39AM +0900, daz wrote:

29

OK, now you lost me. This whole subthread started out at someone
claiming that mails forwarded by rubygate got rejected because of
missing References headers (because they're not following RFC850).

So it's pretty clear (to me at least), that in order for your claim to
be true, rubygate must insert References headers on mails with a
'Re: ' subject which are missing it. Whether it's actually doing that
now, sorta got lost in the noise.

···

On Mon, Jun 07, 2004 at 04:53:40AM +0900, daz wrote:

> > and
> > RFC850: Standard for Interchange of USENET Messages
> >
> > "... the rule is adopted that all USENET news articles
> > must be formatted as valid ARPANET mail messages,
> > according to the ARPANET standard RFC 822."
> >
> > Our missing ML-> NG messages, after conversion from mail to news by rubygate,
> > are valid to RFC822/RFC850 standard regardless of the MUA used for posting.
>
> That's reverse logic, it only claims that all news messages is valid
> mails, not that all mail is valid news.

The claim that ruby-talk mail is converted by rubygate to RFC850 news
comes from me, supported by my (and others') empirical evidence.

--
Thomas
beast@system-tnt.dk

30

Thomas Fini Hansen wrote <before heavy snippage>:

OK, now you lost me.

If this summary helps, please re-read the rest of your message
and see why it would add to the confusion if I tried to
unravel that knot :slight_smile:

···

-----

The reason why some ML->NG messages were/are being dropped is
because the server at news.zrz.tu-berlin.de ("news.zrz") is
rejecting them as being invalid.

As I understand it, David and Dennis were able to reduce the
number of rejections by adding to the headers from within the
g/way software ("rubygate").
David's messages (and some others) are now getting through.
(Dennis had been careful to point out that he hadn't been
expecting 100% success from the change.)

Botp then found that one of his ruby-talk emails didn't get
to c.l.ruby. I noticed another.

             - - -

All the various email apps (Mail User Agents - MUAs) around
produce their own sets of "correct" headers.

The gateway (ML->NG direction) converts mail from ruby-talk
to news by filtering out non-news headers and tries to post to
comp.lang.ruby using NNTP via "news.zrz" (the news host).

ruby-talk ML <--> rubygate <--> "news.zrz" <--> Usenet (c.l.ruby)

My observation is that "news.zrz" also has an opinion on which
combinations of outbound headers are "correct" and that the
criteria for "correctness" are stricter than RFC850 for Usenet.
My news host, for example, will accept the exact same
message that "news.zrz" rejects. It appears on Usenet.

             - - -

To fix the problems, Dennis has suggested that we could try to
find a formal header structure to smooth over the anomalies.

Indeed, it might be possible to adapt the output from the
many MUAs into a normalised form.

My question is: Is this the best way to proceed ?

The manipulation of headers would be to overcome the "news.zrz"
restrictions; not to make them acceptable to Usenet.
(Usenet would accept them from rubygate *without change*).

             - - -

I wouldn't want the gateway to be moved again because the
service at TU-Berlin has appeared to be stable and fast.

At the same time, we can't ask TU-Berlin to change the
software on their news host.

IMHO, we shouldn't be complicating the gateway or ruby-talk
mailer software to satisfy the requirements of our current
service which will not accept all RFC850 news posts.

Proof, from earlier, is on Google, Usenet and ruby-talk archive -
not a wild claim.

--
    daz

If things are less clear, there may be some details in
previous postings on this subject which need to be unlearnt.

31

Hi --

···

On Tue, 8 Jun 2004, daz wrote:

To fix the problems, Dennis has suggested that we could try to
find a formal header structure to smooth over the anomalies.

Indeed, it might be possible to adapt the output from the
many MUAs into a normalised form.

My question is: Is this the best way to proceed ?

I think so, not because it is ideal or non-kludgy but because I want
to see all the messages make it through, and, in practical terms, I
think that's by far the highest-percentage way to make that happen.

David

--
David A. Black
dblack@wobblini.net

32

David A. Black wrote:

Hi --

> To fix the problems, Dennis has suggested that we could try to
> find a formal header structure to smooth over the anomalies.
>
> Indeed, it might be possible to adapt the output from the
> many MUAs into a normalised form.
>
> My question is: Is this the best way to proceed ?

I think so, not because it is ideal or non-kludgy but because I want
to see all the messages make it through, and, in practical terms, I
think that's by far the highest-percentage way to make that happen.

By far the highest-percentage way to make that happen would be
to have a news host that doesn't locally filter outbound traffic :slight_smile:

It would be non-kludgy, require no changes to either g/w or
fml mailer software and allow 100% ML->NG mirroring -- like it
did before the end of January this year.

Do we know that it's not practical to achieve that, yet?

What software change/update was made on that server around Jan 28th ?

What protection is it trying to offer and to whom?
It can't be for security because day after day it's allowing the g/w
to "spoof with implicit permission" NG posts made up from ML members
emails. But if you try posting a reply without a References: header
.... oh, no, sorry, we can't allow that, it might upset someone's
newsreader threading or something... and returns an NNTP error as if
it were an NNTP violation (which it isn't).

Could it have a bug which the authors might be grateful to know about?

Is there a possibility to register the g/w as a privileged user on it?

And other questions that we might ask before coming to a decision
about what would be the best way to proceed.

More cents,

daz

···

On Tue, 8 Jun 2004, daz wrote:

33

Hi --

David A. Black wrote:
> Hi --
>
>
> > To fix the problems, Dennis has suggested that we could try to
> > find a formal header structure to smooth over the anomalies.
> >
> > Indeed, it might be possible to adapt the output from the
> > many MUAs into a normalised form.
> >
> > My question is: Is this the best way to proceed ?
>
> I think so, not because it is ideal or non-kludgy but because I want
> to see all the messages make it through, and, in practical terms, I
> think that's by far the highest-percentage way to make that happen.
>

By far the highest-percentage way to make that happen would be
to have a news host that doesn't locally filter outbound traffic :slight_smile:

It would be non-kludgy, require no changes to either g/w or
fml mailer software and allow 100% ML->NG mirroring -- like it
did before the end of January this year.

Do we know that it's not practical to achieve that, yet?

What software change/update was made on that server around Jan 28th ?

What protection is it trying to offer and to whom?
It can't be for security because day after day it's allowing the g/w
to "spoof with implicit permission" NG posts made up from ML members
emails. But if you try posting a reply without a References: header
.... oh, no, sorry, we can't allow that, it might upset someone's
newsreader threading or something... and returns an NNTP error as if
it were an NNTP violation (which it isn't).

Could it have a bug which the authors might be grateful to know about?

Is there a possibility to register the g/w as a privileged user on it?

And other questions that we might ask before coming to a decision
about what would be the best way to proceed.

All of your ideas and questions are fine. What I'm saing is I'd like
to see something done as quickly as possible -- and, in fact,
something seems to have been done, because I believe Austin Ziegler's
posts are now getting through. If it's just a stopgap measure, so be
it; it's important to stop the gap, and to get as many messages
through as possible while addressing the longer-term or more difficult
questions.

David

···

On Wed, 9 Jun 2004, daz wrote:

> On Tue, 8 Jun 2004, daz wrote:

--
David A. Black
dblack@wobblini.net

34

What protection is it trying to offer and to whom?
It can't be for security because day after day it's allowing the g/w
to "spoof with implicit permission" NG posts made up from ML members
emails. But if you try posting a reply without a References: header
.... oh, no, sorry, we can't allow that, it might upset someone's
newsreader threading or something... and returns an NNTP error as if
it were an NNTP violation (which it isn't).

Doesn't seem much different than a SMTP server returning an SMTP error
when SpamAssassin fires off.

Indeed it's not stricktly a NNTP error, but they're rejecting posts
that violate the USENET RFC, so why doen't we just make the messages
meet the RFC?

Could it have a bug which the authors might be grateful to know about?

Perhaps it's a feature?

And from your other mail:

My observation is that "news.zrz" also has an opinion on which
combinations of outbound headers are "correct" and that the
criteria for "correctness" are stricter than RFC850 for Usenet.

No matter how many times I read RFC 850 section 2.1.6 (or RFC 1036 for
that matter), I cannot see how this should be 'stricter' in any
way. On the contrary, it's criteria seems to be compliance.

My news host, for example, will accept the exact same
message that "news.zrz" rejects. It appears on Usenet.

That just proves that your news host has more lax requirements, not
that the article is in fact valid. To pull the SMTP comparison again,
our mailserver routinely reject mails that violate the specs, yet
another mailserver didn't reject them.

Frankly, I can't see what the big fuzz is about, inject the damn
References header, it's the right thing to do, whether one considers
it 'klunky' or not.

···

On Wed, Jun 09, 2004 at 03:53:44AM +0900, daz wrote:

--
Thomas
beast@system-tnt.dk

35

Thomas Fini Hansen wrote:

> newsreader threading or something... and returns an NNTP error as if
> it were an NNTP violation (which it isn't).

Doesn't seem much different than a SMTP server returning an SMTP error
when SpamAssassin fires off.

SMTP error or SMTP USER error ?
Does a SpamAssassin rejection get reported as a badly formatted message ?
BTW: I don't need to know.

Indeed it's not stricktly a NNTP error, but they're rejecting posts
that violate the USENET RFC, so why doen't we just make the messages
meet the RFC?

THE MISSING MESSAGES DO *NOT* VIOLATE USENET RFC.

The *local* news host that we're using doesn't like them
and won't forward them to Usenet.

> Could it have a bug which the authors might be grateful to know about?

Perhaps it's a feature?

I think you're being contrary.
It /could/ be a feature, but it's not a very useful feature to us, is it?
There could be other services using that news host which are getting
their posts dumped and are not aware of it.

And from your other mail:

> My observation is that "news.zrz" also has an opinion on which
> combinations of outbound headers are "correct" and that the
> criteria for "correctness" are stricter than RFC850 for Usenet.

No matter how many times I read RFC 850 section 2.1.6 (or RFC 1036 for
that matter), I cannot see how this should be 'stricter' in any
way. On the contrary, it's criteria seems to be compliance.

Our messages are already compliant !
"news.zrz" is stricter in that it insists on an RFC-optional header (References:).

> My news host, for example, will accept the exact same
> message that "news.zrz" rejects. It appears on Usenet.

That just proves that your news host has more lax requirements, not
that the article is in fact valid.

What proves validity is that it arrived on Google.
What proof are _you_ looking for ??

To pull the SMTP comparison again {...}

It's unhelpful. We can talk about NNTP.

Frankly, I can't see what the big fuzz is about, inject the damn
References header, it's the right thing to do, whether one considers
it 'klunky' or not.

That's been done.

--
Thomas
beast@system-tnt.dk

If I'm wrong about this, I'm sure someone else will let me know.

daz

···

On Wed, Jun 09, 2004 at 03:53:44AM +0900, daz wrote:

36

Hi --

THE MISSING MESSAGES DO *NOT* VIOLATE USENET RFC.

Here are some excerpts from RFC-1036 (which obsoleted RFC-850):

  2.1.4. Subject

  [...] If the message is submitted in response to another message
  (e.g., is a follow-up) the default subject should begin with the
  four characters "Re: ", and the "References" line is required.

  [...]

  2.2.5 References

  [...] User interfaces need not make use of this header, but all
  automatically generated follow-ups should generate the "References"
  line for the benefit of systems that do use it, and manually
  generated follow-ups (e.g., typed in well after the original message
  has been printed by the machine) should be encouraged to include
  them as well.

Based on this, I would definitely say that "Re: " messages without a
References header are in violation of this RFC.

I think the confusing thing is that the RFCs list References among
"optional headers", when it becomes clear from the more detailed
descriptions above that what they mean by "optional", at least for
this header, is "required for some, but not all, messages". (In other
cases I think it means "completely discretionary", as for example with
Organization.)

David

···

On Thu, 10 Jun 2004, daz wrote:

--
David A. Black
dblack@wobblini.net

37

David A. Black wrote:

> THE MISSING MESSAGES DO *NOT* VIOLATE USENET RFC.

Here are some excerpts from RFC-1036 (which obsoleted RFC-850):

  2.1.4. Subject

  [...] If the message is submitted in response to another message
  (e.g., is a follow-up) the default subject should begin with the
  four characters "Re: ", and the "References" line is required.

  [...]

  2.2.5 References

  [...] User interfaces need not make use of this header, but all
  automatically generated follow-ups should generate the "References"
  line for the benefit of systems that do use it, and manually
  generated follow-ups (e.g., typed in well after the original message
  has been printed by the machine) should be encouraged to include
  them as well.

Based on this, I would definitely say that "Re: " messages without a
References header are in violation of this RFC.

I think the confusing thing is that the RFCs list References among
"optional headers", when it becomes clear from the more detailed
descriptions above that what they mean by "optional", at least for
this header, is "required for some, but not all, messages". (In other
cases I think it means "completely discretionary", as for example with
Organization.)

David

"Request For Comments" ? Makes me wonder if anyone did :slight_smile:
                                (comment, I mean)

Your interpretation, even if it's not correct, should be, IMHO.

References should be listed as "conditional", not "optional" ?

With that small revision, I could withdraw my CAPS-LOCKed statement.

Usenet itself doesn't consider them to be invalid. If "news.zrz"
were to send regardless, they would make it around the Usenet
servers.

···

On Thu, 10 Jun 2004, daz wrote:

-----

According to http://www.ietf.org/iesg/1rfc_index.txt (long),
850 & 1036 are both Status:UNKNOWN so perhaps there is no standard :slight_smile:

   0850 Standard for interchange of USENET messages. M.R. Horton.
        Jun-01-1983. (Format: TXT=43871 bytes) (Obsoleted by RFC1036)
        (Status: UNKNOWN)

   1036 Standard for interchange of USENET messages. M.R. Horton, R.
        Adams. Dec-01-1987. (Format: TXT=46891 bytes) (Obsoletes RFC0850)
        (Status: UNKNOWN)

-----

daz

38

Hi --

David A. Black wrote:
>
> Here are some excerpts from RFC-1036 (which obsoleted RFC-850):
>
>
> 2.1.4. Subject
>
> [...] If the message is submitted in response to another message
> (e.g., is a follow-up) the default subject should begin with the
> four characters "Re: ", and the "References" line is required.
>
> [...]
>
> 2.2.5 References
>
> [...] User interfaces need not make use of this header, but all
> automatically generated follow-ups should generate the "References"
> line for the benefit of systems that do use it, and manually
> generated follow-ups (e.g., typed in well after the original message
> has been printed by the machine) should be encouraged to include
> them as well.

[...]

Usenet itself doesn't consider them to be invalid. If "news.zrz"
were to send regardless, they would make it around the Usenet
servers.

In practical terms, though, the spelling out of the requirement in the
RFC probably means that admins who have decided to comply with it are
not likely to be talked out of it, so there's probably not much point
pursuing that particular path. I'm not sure what that leaves us with.
Not much, I fear. I was thinking about maybe having a dummy message,
to which all these stranded messages could be sent as replies. It
would presumably take them out of the threads they were supposed to be
in, but at least it would get them through.

David

···

On Thu, 10 Jun 2004, daz wrote:

--
David A. Black
dblack@wobblini.net