[net.unix-wizards] Incorrect "References:" fields in news items

michaelr@elecvax.OZ (Michael Rourke) (06/15/84)

I have implemented my own news software, and thought it would
be nice to be able to unsubscribe from particular followups
using the "References:" fields. Mark Horton's article "Standard
for Interchange of USENET Messages" defines how the "References:"
field should be generated. Trouble is most news software around
gets it wrong!

This is most annoying.

To illustrate I have collected some headers from a particular discussion:

The first two seem OK (I don't have the original article):

  From: merlyn@sequent.UUCP
  Subject: Re: Ethernet addressing (and destroying dollar bills)
  Message-ID: <515@sequent.UUCP>
  References: <1329@uvacs.UUCP> <1429@cbosgd.UUCP>
  
  From: faunt@saturn.UUCP
  Subject: Re: Ethernet addressing
  Message-ID: <2138@saturn.UUCP>
  References: <1329@uvacs.UUCP> <1429@cbosgd.UUCP>
  
This is wrong:

  From: dgary@ecsvax.UUCP
  Subject: Re: Ethernet addressing (and destroying dollar bills)
  Message-ID: <2678@ecsvax.UUCP>
  References: sequent.515

Not only is the reference field the wrong format, but the software
didn't copy the previous references field from the message.
  
This is interesting too:

  From: jaap@haring.UUCP
  Subject: Re: Ethernet addressing
  Message-ID: <260@haring.UUCP>
  References: <999@hao.UUCP>
  
Wrong again (at least the field format is correct):

  From: ed@mtxinu.UUCP
  Subject: Re: Ethernet addressing (and destroying dollar bills)
  Message-ID: <128@mtxinu.UUCP>
  References: <515@sequent.UUCP>
  
And now we have a comma in there: (again non-standard)

  From: karn@mouton.UUCP
  Subject: Re: Ethernet addressing (and destroying dollar bills)
  Message-ID: <76@mouton.UUCP>
  References: sequent.515, <2678@ecsvax.UUCP> <651@pyuxhh.UUCP>
  
And so on....

  From: kurt@pyuxhh.UUCP
  Subject: Re: Ethernet addressing (and destroying dollar bills)
  Message-ID: <651@pyuxhh.UUCP>
  References: sequent.515, <2678@ecsvax.UUCP>
  
  From: rpw3@fortune.UUCP
  Subject: Re: Ethernet addressing - (nf)
  Message-ID: <3564@fortune.UUCP>
  
This one is correct - we generated it here:

  From: dave@elecvax.SUN (Dave Horsfall)
  Subject: Re: Ethernet addressing
  Message-ID: <232@elecvax.SUN>
  References: <1329@uvacs.UUCP> <1429@cbosgd.UUCP> <2138@saturn.UUCP>

I don't expect everybody to rush out and fix their source(s),
but it would be nice if further releases of the news software:

	1. provided an "unsubscribe from followups" command
	2. generated the References: field as per the standard

Ways of filtering large volumes of news are useful.


Michael Rourke
Department of Computer Science
University of New South Wales, Australia

..(decvax,vax135,sfjec)!mulga!michaelr:elecvax

mp@whuxle.UUCP (Mark Plotnick) (06/22/84)

Here are some fixes for the bogus References lines.

- In followup_command (and reply_command), change the code that
concatenates ", " to use " " instead.

- the code in hbufcp is kind of old, and doesn't copy the followid
field.  So, when a user does an f- command (the usual state of affairs),
there's no References line to be found in the followed-up-to article!
Hbufcp should really do a structure copy (except for the unrec
fields).

- the code in followup_command (and reply_command) blindly does lots of
strcats without looking to see whether it's overflowing folbuf.  If the
References line gets too long, folbuf won't be null-terminated, and the
string starting at folbuf will continue into the subj array (on a vax,
at least).  That's why you'll sometimes see "Subject:" near the end of
a References line.  The fix is to either have the code do strncats or
else just make folbuf[] immensely huge.

- the illegal Message-IDs are probably a result of the pre-2.10 news
systems that are still on the net.

	Mark Plotnick