From: J.B.Moreno (planb@newsreaders.com)
Date: Mon Mar 15 2004 - 00:10:14 CST
On 3/14/04 4:03 PM, Bruce Lilly at <blilly@erols.com> wrote:
> J.B.Moreno wrote:
>> On 3/12/04 9:31 AM, Bruce Lilly at <blilly@erols.com> wrote:
>>
>>
>>> J.B.Moreno wrote:
>>>
>>>
>>>> That's not my point of disagreement -- I really do think that the "Re: " is
>>>> part of the syntax...
>>>
>>> Can you substantiate that in engineering terms based on RFC 822 syntax
>>> (text and EBNF), which has been the basis for Usenet articles for at least
>>> two decades, or based on RFC 2822 syntax (text and ABNF) which is currently
>>> the basis for the Usefor draft?
>>
>> Are you asking me to point out where they defined the syntax to mean that?
>> If so, they don't, at least not explicitly. If that's your question, then
>> my answer is "the made a mistake, we shouldn't make the same mistake".
>
> That still lacks substance. Moreover, it doesn't address the fact of
> prior use, dating back to RFC 561. The message header field name
> registry is about to become a reality, and that very strongly discourages
> the same field name from being used in different protocols with
> different meanings. Now if you want to go back to "Title",...
Prior usage?!?
You are aruging prior usage? When followingup the Subject is prefixed with
"Re: " -- that is prior usage. Goes back to at least '81. That's a lot of
prior usage.
As for the registry -- fine, everyone else can use our definition, it's what
they *do* even if it's not how they define it.
>>>> Saying that the Subject is only for display, isn't true
>>>
>>> Please provide an example of use unrelated to display. Show how it affects
>>> filing, transfer, expiration, etc. of the article.
>>
>>
>> Filtering determines whether something is filed or not. Spam filtering
>> determines whether it is transfered or not. I know that some of the
>> webboards with NNTP interfaces don't expire FAQs and may determine this in
>> part based upon the Subject.
>
> And use of Subject (or Date, etc.) for any of those purposes is no
> more relevant to the string "Re" than to the string "\/|4%r4" --
> probably a great deal less so. You still haven't made a case that
> "Re" is in any way shape or form part of the syntax or protocol.
It's part of saying that it's a followup. It has a defined meaning (that
the Subject wasn't created fresh just for this post). It's presence is nigh
universal, it has a precise position in the Subject, it's detectable and in
practice (not theory) it *is* less error prone than the References header.
How one could think that it's not part of the syntax is beyond me. Your
argument against it being part of the syntax seems to boil down to 2822 says
the Subject can't have a syntax.
>> I'm saying that the point of a message is to be read, a message that isn't
>> read might as well not have been sent. It doesn't matter if it's not read
>> because it got misfiled or expired immediately or because it grabbed the
>> Subject from an article 3 groups over. If a newsreader author follows our
>> rules, and as a result of doing so, the messages his clients send out,
>> aren't read when by ignoring our rules they would have been, then our rules
>> should be ignored.
>
> And that is affected by "Re" ... how?
The decision to read an article or not is sometimes based upon the correct
usage of "Re: " -- so we need to tell the authors what is correct, so that
they can have their clients behave correctly.
-snip-
>>>> The "Re: " contains information -- it says that the Subject was inherited
>>>> (give or take some minor changes)
>>>
>>> It does not contain the information that you claim that it contains:
>>> Subject: Re: is an abominable hack
>>
>> What does that have to do with anything? Possibly it's blasphemous as well,
>> it's there, it's used, you don't like it, tough.
>
> The example Subject field line was *NOT* "inherited" -- it appears in an
> article which is not a followup. Ergo your claim about what "Re" means
> is false.
Infailibility is *NOT* a requirement of any header field.
If I post a new message and it has a References header, that doesn't mean it
was in fact a followup -- it means that either I (the poster) or my
software, didn't follow the rules, and as a result other people will *think*
that it was a followup.
>>> appearing in an article which is not a followup most certainly does *NOT*
>>> "say[...] that the Subject was inherited" or anything remotely like that.
>>
>> Sure it does. It can be mistaken, but so what -- the References header can
>> be mistaken as well. That just means someone or something screwed up, it
>> doesn't mean that it's meaningless.
>
> Well not to put too fine a point on it, but the guy who *assumed*
> (incorrectly) that "'Re" says that the Subject was inherited" is
> the someone who screwed up. An incorrect assumption isn't
> "information" -- it might well be "mistaken", but that's another
> matter entirely.
And if I write a newsreader that creates a References header for ever post,
using a random message-id from a random group -- does that make the software
that "assume" that the posts made with my newsreader are followups to those
random articles mistaken?
The fact of the matter is, "Re: " is less error prone than the References
header is in *practice* as opposed to theory.
-snip-
>>> So this non-existent information (really a misinterpretation) is (ab)used --
>>> for what (other than display or display-related processing) you haven't been
>>> able to say. And somehow this "information" gets from the followup agent to
>>> some other reading agent even though the poster has elided it (which he is
>>> explicitly permitted to do)? WTF?
>>
>> If he removes it, then *that* information is conveyed!!!!
>
> But you can't tell whether it was never there or if it was removed!
Only in the sense that you can't tell a truth from a lie, or corrupted data
from uncorrupted without verifying the facts.
Absent bad software or user intervention, a Subject that doesn't start with
"Re: " was created specifically to refer to the contents of that post, a
Subject that *does* start with "Re: " wasn't, it was in whole or in part
inherited from the article of which it is a followup.
Just as absent bad software or user intervention, the last message-id in the
References header was inherited from the Message-ID header of the article
for which this article was a followup.
And in practice, as opposed to theory, the Subject is less often "mistaken"
for either reason.
>> The poster can really only "remove" the information by putting in "Re: "
>> when he shouldn't
>
> If the presence of "Re" conveys some "information", and that "Re" is
> removed, then presumably that "information" is removed (without
> "putting in" anything).
The "Re: " is a negative symbol -- it's removal creates a positive, not
nothing.
-- J.B. Moreno