AW: AW: -- Whats this 'AW:' mean?

Scott A Crosby crosby at qwes.math.cmu.edu
Sat Feb 2 11:43:27 UTC 2002


On 2 Feb 2002, Cees de Groot wrote:

> (did that sound sarcastic? Well, it was meant to. Go to large parts of the
> worlds, and people will *not* understand English and - justifiably - have no

Who said anything about English. its from 'in re', which is Latin!

Blame latin speakers, not english speakers.
  http://www.faqs.org/usefor/1997/Jul/0134.html

So, find the nearest latin speaker and bitch them out, not me.


> sympathy at all with such views. I fail to see why Germans should generally

Because when all clients start adding random prefixes...... Two more
round-trips with this email and the subject will be cut off on the right
side of the window.

Adding random prefixes make the subject header useless to for humans --
hard to parse and read. And computers -- it will break naieve clients who
do cheap threading by sorting on subject&date (after canonicalizing by
removing any 're:' prefix.) This works pretty well, and there are a lot of
naive clients... Is Celeste one of them?

> use something they don't understand when they send mails just in case that
> someone somewhere in the USofA is inconvenienced because his mailreader
> doesn't follow in-reply-to when threading.... Really, everyone in Germany

Not robust. With a sampling of the messages on the list, barely half have
any In-Reply-To or References header. Those that do only mostly have one
such email, the one they're replying to. Thus, for each missing
(lost/deleted) message in some archive, the parentage tree will gain at
least one partitian per missing email. By partitian, I mean there's no way
to calculate that two threads which are siblings actually are siblings.

And, a fallback of 'sort on subject' will be thwarted by the randomly
added prefixes on the subject lines.

> understands 'Antwort'. If there's one thing to complain about, it is that the
> related RFC's are lacking here which apparently necessitates the Re/AW/...
> thingies in Subject: lines).

Any sophisticated client that uses References/In-Reply-To can't deal with
missing data. Adding random prefixes makes the backup naive approaches
they use fail. (Try this out experimentally with your mail spool)

Eventually all clients will thread by references, and will be required to
preserve the entire reference chain to the first email. [*] This'll
probably take 5-10 years to go through all the clients everywhere.

But, in the mean time, each language's client will be clobbering the
subject line with random prefixes, and making the 'Subject' of email
useless and unreadible. Then, the RFC's will notice this stupidity and
standardize on a new 'Human-Subject' header that is guarenteed to never be
altered by machine..

And the total gain of this, over a sort-by-subject-and-date is... what?

Well, I would appreciate a good threading mail reader.

But, I'd not want the pain, broken clients, and hard to follow/read email
for the next decade, because some people dislike the latin 're:' prefix
for replies?

Scott

[*] There are interesting edge-cases in the reconstruction algorithm,
especially when the data you feed it is mutually inconsistent. A malicious
person forging these References lines could certainly produce some
fun. :)




More information about the Squeak-dev mailing list