🚀 go-pugleaf

On Friday, in article <852259918snz@tsys.demon.co.uk>
     tw104@york.ac.uk "Tom Wheeley" wrote:

TW> I don't know how KA9Q operates exactly, but I suspect the best way would be
TW> to modify it to bin completely articles with headers matching entries in the
TW> killfile prefixed with something like '+'.

?  As in 'add this'... NOT!  However, nit-picking over the character
to use is probably the least difficult (and therefore will receive the
most discussion <g>).

The usual cry at this point (*) is that memory is so tight in ka9q.
But, looking at the similarity of the killlist/keeplist logic
in nntpcli.c, perhaps a bit of smart coding might even make
this doable in 0 bytes!

Perhaps the most error-prone part of the addition would be
the logic to omit the header.  At present I assume the header
is being added as it's being received and scanned.  Maybe
the simplest way of extending this would be to note the
location of the start of the header and then, when a
kill-and-bury match is seen, truncate back to there.

I certainly agree it sounds more appealing than having
a separate utility to post-process the file.

(*) By DOS-ka9q developers, I mean.  In view of the cross-posting
I have to add I wouldn't even know if OS/2 has broken through
the 640K barrier.  :-)

--
John Washington