[news.software.nn] sorting of subject line

ccastjb@prism.gatech.EDU (Joubert Berger) (01/24/90)

I don't know if I missed something, but anyway here goes...

Nn is supposed to sort the subject lines.  That works for me.  Now after
all subjects have have been sorted and there are a few messages with the
the same subject heading, what does nn do?  I sometimes find that I am reading
a reply to a message before I read the original message.  Is this a bug, or
did I miss something.  
                 Joubert
-- 
Joubert Berger                    | uucp:  ...!{allegra,amd,hplabs,seismo,
Georgia Institute of Technology,  |             ut-ngp}!gatech!prism!ccastjb
P.O.Box 36152                     | ARPA:  ccastjb@prism.gatech.edu
Atlanta Georgia, 30332            | 

rcodi@koel.co.rmit.oz (Ian Donaldson) (01/24/90)

ccastjb@prism.gatech.EDU (Joubert Berger) writes:
>Nn is supposed to sort the subject lines.  That works for me.  Now after
>all subjects have have been sorted and there are a few messages with the
>the same subject heading, what does nn do?  I sometimes find that I am reading
>a reply to a message before I read the original message.  Is this a bug, or
>did I miss something.  

It is probably a side effect of the huge number of people that don't
know how to set the correct time on their machine.  They set the
local-time correctly (mostly) but don't bother setting the timezone
correctly before hand.  Generally the timezone is the same as
the one the manufacturer left on the distribution tapes.

The result is that the GMT value of time on their machine is incorrect,
and it is this value that appears in "Date:" headers in articles,
upon which nn sorts the order.

Ian D

davison@drivax.UUCP (Wayne Davison) (01/25/90)

Joubert Berger (ccastjb@prism.gatech.EDU) wrote:
} Nn is supposed to sort the subject lines.  That works for me.  Now after
} all subjects have have been sorted and there are a few messages with the
} the same subject heading, what does nn do?  I sometimes find that I am reading
} a reply to a message before I read the original message.  Is this a bug, or
} did I miss something[?]

There are a few possibilities (besides the GMT dates being in error):

Are the dates you are reading of the C news style? (with a leading day-of-the
week? E.g. Wed, 24 Jan 90 23:59:59).  If so, you need a fairly recently patched
version of nn (patchlevel 6 works) to read the dates.  Older versions parsed
all such dates into zero, which sort in an arbitrary order.

Another possibility is that someone changed the subject of the reply article
to one that was alphabetically prior to the parent article.  In this case you
read the reply first because it is in a lower group alphabetically, and then
read the parent article later on in the session.  I sent some code to Kim Storm
last November (or so) that corrects this problem by first sorting the articles
into their subject groups (like the present method) and THEN sorts the groups
by date of their first article.  This way nn shows you the older articles first
no matter what their subjects turn into.  You can still see the alphabetical
sort with the command ":sort subject".  If people are interested, I could post
the simple patch.
-- 
Wayne Davison          \  /| / /| \/ /| /(_)         davison@drivax.UUCP
                      (_)/ |/ /\| / / |/  \          ...!amdahl!drivax!davison

king@batserver.cs.uq.oz.au (Paul King) (01/26/90)

rcodi@koel.co.rmit.oz (Ian Donaldson) writes:

>ccastjb@prism.gatech.EDU (Joubert Berger) writes:
>>Nn is supposed to sort the subject lines.  That works for me.  Now after
>>all subjects have have been sorted and there are a few messages with the
>>the same subject heading, what does nn do?  I sometimes find that I am reading
>>a reply to a message before I read the original message.  Is this a bug, or
>>did I miss something.  

>It is probably a side effect of the huge number of people that don't
>know how to set the correct time on their machine.  They set the
>local-time correctly (mostly) but don't bother setting the timezone
>correctly before hand.  Generally the timezone is the same as
>the one the manufacturer left on the distribution tapes.

>The result is that the GMT value of time on their machine is incorrect,
>and it is this value that appears in "Date:" headers in articles,
>upon which nn sorts the order.

>Ian D

I get this effect from people who change the subject line
i.e., don't use the followup features correctly.

For instance, if I was to follow up to this article but change
the subject line to `correct reason for sorting foulups'
instead of `Re: sorting of subject line' then alpabetically
my reply would come before the original.  So depending on
how often you read your news and where abouts in the world
you are in relation to the sender(s), you are likely to
read my followup before the original.  Luckily, I use the
followup option built-in to nn and am to lazy to even
contemplate finding out if it is possible to change the
subject line, so most people won't get out of order messages
from me.

Paul K

moraes@cs.toronto.edu (Mark Moraes) (01/26/90)

king@batserver.cs.uq.oz.au (Paul King) writes:
>I get this effect from people who change the subject line
>i.e., don't use the followup features correctly.

Changing the subject line IS correct if you change it to match the
text of the article.  It does defeat newsreaders that believe subject
== thread, and kill on subject, so it isn't worth doing unless the
discussion has really drifted, which Usenet discussions tend to do.
Wayne Davison's idea to sort subject groups by date is a nice one.

Using the References: header to sort would be nice too, though not as
simple/fast.  This is intertwined with the threaded newsreader thread
in news.software.b, but not quite the same -- I'd be happy with using
the References: for a simple topological sort.  Excuses that this
scheme can get tied up in knots because some news software mangles the
References: line aren't really valid; such mangling is a bug, and
should be fixed -- the mangling can be ignored (it typically consists
of chopping off enough characters from the last message-id to fit a
new one in).  And of course, there are certain SMTP agents that refuse
to accept news by mail because the references line is longer than 254
characters, so news<->mail gateways would have to split the line
carefully, and glue it back together.  And there are certain fussy
mailers that actually validate every message-id in the references line
to make sure it conforms to RFC822 syntax, and chide the sender with
pedantic error messages....