[Users] The reply mark in subject (R: Re:) recognition

[Holger Berndt]

On Do, 09.08.2012 12:08, Abhay S. Kushwaha wrote:

> Holger, please allow me to break it down in a more structured manner.

Absolutely! I've seen so many (false) claims backed up with a fuzzy
reference to an RFC over the years that I really appreciate somebody
going down to the spec text. I still don't completely agree with you,
but at least now I have a concise interpretation to disagree with ;)

I'll inline in both, your questions and answers.

> 1. Is "Antwort: " allowed to be the "Subject:" field?
> 2. Should a response to an email be identified by modifying the
>    "Subject:" field?
> 3. If the answer to #2 allows modification of "Subject:" field, is
>    there a prescribed way to do it?

Stop. In #2 you didn't ask whether modifications of the "Subject:" field
are _allowed_, but if they _should_ be done.

> 4. If there is a prescription in #3, are "Antwort: " or "Aw: " (its
>    abbreviation) part of that prescription?
> 
> Now the answers.
> 
> 1. Yes, of course. Any text is allowed to be in "Subject:" field. In
>    that, the used text is the subject of the email itself.
> 
>    Ref: 3.6.5, first paragraph.
> 
>    > The "Subject:" and "Comments:" fields are unstructured fields
>    > as defined in section 2.2.1, and therefore may contain text
>    > or folding white space.

Agreed.

> 2. It is allowed that a reply to an email message be indicated by
>    modification of the "Subject:" field. It is not compulsory to do
>    so, but it is allowed to be done.

Agreed on the claim, not on the RFC reference supporting it.

>    Ref: 3.6.5, second paragraph.
> 
>    > The "Subject:" field is the most common and contains a short
>    > string identifying the topic of the message.  When used in a
>    > reply, the field body MAY start with the string "Re: " (from the
>    > Latin "res", in the matter of) followed by the contents of the
>    > "Subject:" field body of the original message.

It is allowed not because of this optional feature description, but
because of the very definition of an "unstructured" header field:

	Ref 2.2.1
	Some field bodies in this standard are defined simply as
	"unstructured" (which is specified below as any US-ASCII
	characters, except for CR and LF) with no further restrictions.

Note that no further restrictions apply. In particular, it is not
mandatory that a reply keeps the "Subject:" header in the first place!

> 3. Yes. If it is choosen that the reply be identified through the
>    modification of the "Subject:" field, then it must and only must
>    use the string "Re: " and no other, and that too, only once and
>    not more than once.
> 
>    Ref: 3.6.5
> 
>    > If this is done, only one instance of the literal string "Re: "
>    > ought to be used since use of other strings or more than one
>    > instance can lead to undesirable consequences.

You're mixing two things here. The first part is what modifications
to the "Subject:" field are allowed. It seems that we agreed that
anything goes (your #1, my addendum to #2).

The second part is whether this modification is supposed to be used
as an identification hint for the receiver of the reply. I agree
that if a client is using "AW: " instead of "Re: ", this optional
feature is not used. That's perfectly fine, though. A client doesn't
need to use this feature. That's why it's optional. The real,
machine-readble reply marker for a mail message is defined elsewhere
anyways (In-Reply-To, References - note that those are "structured",
as opposed to "unstructured", for a reason).

> 4. No. The prescribed format of identifying a response through
>    modification of "Subject:" field should use only use "Re: ", and
>    only once, and no other string or multiple uses of "Re: " are
>    prescribed.
> 
>    Ref: 3.6.5
> 
>    > If this is done, only one instance of the literal string "Re: "
>    > ought to be used since use of other strings or more than one
>    > instance can lead to undesirable consequences.
> 
>    Ref: A.2
> 
>    > When sending replies, the Subject field is often retained, though
>    > prepended with "Re: " as described in section 3.6.5.

Correct, if you want to implement the optional feature. Again - it's
not an RFC violation not to do that!

In particular,
  Subject: Re: Re: old subject
is not an RFC violation - it's just not the feature described in 3.6.5.

> So when you use any other string than "Re: " in an attempt to
> identify a response, you should not expect that a software or human
> who correctly understands and implements the RFC will understand.

Indeed. However, I never claimed that.

> In other words, if I receive an email with:
>   Subject: Aw: Hi Holger
> the RFC compliant responses to that email would only be two
>   Subject: Aw: Hi Holger
>    -or-
>   Subject: Re: Aw: Hi Holger
> where, in the first instance I chose not to identify a reply by
> modifying the "Subject:" field and in the second instance I chose to
> identify a reply by modifying the "Subject:" field.

Wrong. Didn't we just agree that anything goes in the subject field?
	Subject: Re: Hi Holger
is perfectly valid. As are all other imaginable "Subject:" fields
(constraint to US-ASCII, no CR/LF).

By the way, if it wasn't, Claws Mail would be the violator here, not
MS Outlook. Ups.

> That is, "Aw: " is not an alternative to "Re: " far as RFC is
> concerned.

True (except the "That is", because it doesn't follow from the
paragraph above).

> The RFC specifically talks of "use of other strings" other
> than "Re: " as potential agents of "undesirable consequences".

The situation is basically as follows:

- The "Subject:" header field can contain anything.
  In particular, it's not an RFC violation if a message contains
  "AW: old subject". That was the claim that I questioned, and so far I
  haven't seen anything that would make me change my mind.

- If the "Subject:" field does not implement the optional feature in
  3.6.5, the receiving MUA might not be able to strip reply indicators
  automatically. Too bad. That's the price you have to pay for making
  it optional.

Now, the $64.000 question is what a receiving MUA should do in case the
optional feature is not being used by the sending MUA. Again, it's not
mandatory that a reply has a certain "Subject:" format. So, Claws Mail
decided to strip known reply marker prefixes other than "Re: " in a way
analogous to section 3.6.5. That seems reasonable to me, and is
certainly allowed.

Holger