[Pan-devel] Re: pan-0.112 how does handle sporges with identical message ID

[Duncan]

Konrad Karl <address@hidden> posted
address@hidden, excerpted below, on  Fri, 29
Sep 2006 15:36:41 +0200:

> In the past (using perl scripts for download) I used to filter on ratio
> bytecount/linecount which is much lower on sporged articles or on the
> senders path and is very likely more than 10 or even 15 on typical
> articles and it worked fully automatic.
> 
> So my question: is the new pan able to address individual articles within
> a binary set in order to delete sporged ones because i cannot see how the
> article number is being recorded besides the first article in a binary
> set, moreover a tcpdump trace showed me, that the first article was
> fetched by article number and the others by Mesage-ID...

pan doesn't at this point do well in scoring/filtering anything where the
poster is deliberately making it difficult to do so, period. 
Unfortunately, it seems the scoring pan does can only handle headers
normally found in the overview, which simply isn't enough to kill these
things.  Better handling has been on the wish list for a very long time. I
know, as I filed a bug on it years ago, but Charles said low priority,
bluesky, likely post 1.0, which was then bluesky as well, and I can't code
or it'd be handled by now.

Well, 1.0 is due shortly as the next stable release (Charles expected it
by now but there were some extra bugs to find and squash), so bluesky
might be on the way as well. =8^)  I suspect 1.1 will add some other
features, automated scored file handling (deletion of killed,
auto-mark-read of below-zero scored, auto-download of watched) likely
among them as old-pan (<0.15, C based while 0.90+ are a C++ based rewrite)
had such features available via rules.  However, I don't personally expect
something major like a reworking of scoring to make it until post-1.1, so
pre-1.2 betas at the earliest.  Breakup of multi-post parts for separate
processing is also post-1.0 now, as that'd be far to large a change to
make it for 1.0.  However, I can't say beyond that.  It'd be up to
Charles, of course.

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman