Sender: jwz@jwz.org
Message-ID: <36D33C9E.19A29117@mozilla.org>
Date: Tue, 23 Feb 1999 15:41:18 -0800
From: Jamie Zawinski <jwz@mozilla.org>
Organization: mozilla dot org
MIME-Version: 1.0
To: drums@cs.utk.edu
Subject: Re: References and In-Reply-To (section 3.6.4)
References: <19990224000248.005992@relay.skynet.be>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Brad Knowles wrote:
> 
>     The problem is, news and mail have always been very similar but
> somewhat distinct, and from what I have seen both here and on USEFOR,
> that distinction seems to be likely to remain forever, whether we like it
> or not.

This is probably true.  But not for good reasons, as far as I can tell.
The reasons seem to me to be primarily that the group of people who care
about mail and the group of people who care about news don't talk to
each other very much, and are each very focussed on their particular
applications.  There are a lot of people who like to focus on the
differences between mail and news as an excuse to ignore the
similarities.  (But I'm not bitter.)

>     I think it would take a *very* concerted effort to consolidate the
> two sets of standards, and both sides would have to accept a lot of stuff
> that they don't like (or don't care about), in order to get the broadest
> possible standard that could encompass both of them.

I'm not talking about consolidating the two sets of standards.
I'm only talking about them interpreting the References header
in the same way!

If even something as simple as that isn't possible, then we should
all just go home.

> That would mean that USEFOR would have to accept messages with
> multiple parents, for example.

I don't accept that multiple parents are useful in mail but not 
useful in news.  If they're useful in one, they're useful in the
other, and vice versa.

I accept that multiple parents might be useful someday, with some
hypothetical as-yet-uninvented user interface.

I don't see what "mail" versus "news" has to do with this.

>     Just your own personal mail, or that which belongs to multiple
> people?  How many total messages is this?  Over what period of time does
> this stretch?

My own personal mail, going back several years; plus some public mailing
list archives I pulled off of the net (but I don't remember which, it
was a long time ago and I neglected to keep track of that.)

>     As kre pointed out, this is not a proper sample.  

I *know* it's not an unbiased sample.
I *said* it's not an unbiased sample.

It is, however, the sample I have.

If you have a better one, I'd love to see it.

> You have pre-selected content from places that are not likely to have
> the sort of thing you were searching for, and therefore you didn't
> find it.  You now declare that it can't possibly exist.

I said no such thing.  I said that in my experience, and in the samples
I searched, I didn't see any usages of multiple parents.  (In fact, I
misspoke when I said that -- my search turned up exactly 4.)

And, by the way, when I did this search last year, I wasn't "searching"
for anything, I merely wanted to understand how In-Reply-To was used in
the real world.  This was as close to "the real world" as I could easily
get.  I just wanted to know what kind of hoops I was going to have to
jump through to (practically) use In-Reply-To for threading in the
absence of a sane References header.

>     However, after getting the appropriate approval from my boss, I might
> be able to do a search of all the existing mailboxes and look for both
> "References:" and "In-Reply-To:" headers and see what kinds of numbers I
> can dig up.

That would be wonderful.

I'm perfectly willing to listen to other evidence.  In the messages I've
examined, multi-parented messages by and large didn't exist, let alone
seem useful.  If you want to argue that they are used and are useful,
then come up with your own data refuting that.  Just telling me that
what I've seen isn't applicable is hardly a productive approach.

If someone else wants to re-write the text about References and
In-Reply-To construction again, feel free.  If the rest of you want
to just revert that text to what RFC 822 said, feel free.  But the
current DRUMS text (which is different from the RFC 822 text) is
pretty bad, and should be replaced once way or another, either by
moving forward or backward.

I've tried to rewrite it twice now and been shot down, it's someone
else's turn.

-- 
Jamie Zawinski
jwz@mozilla.org    http://www.mozilla.org/   (work)
jwz@jwz.org        http://www.jwz.org/       (play)