[Pan-users] Re: Filtering on the basis of a missing header

[Duncan]

JCA <address@hidden> posted
address@hidden, excerpted
below, on  Sat, 30 Jun 2007 14:59:14 -0700:

>      Well, thanks for the info. That does not change the fact that I'd
> still be interested in filtering on the basis of a missing field in the
> header. Do you know how to do so?

I had a reply to this but couldn't get it out due to the 441s on gmane.  
They were gone earlier, let's see if I can still get replies out, and if 
I can find that reply...


I'm not /sure/ this will work, but it /should/ work, tho /only/ for 
headers normally in the overview.  I'd suggest trying it first with a 
score such as 3657 or something equally strange, to ensure it's working, 
before you try setting ignore.

For the header in question, set it (directly in the scorefile) to:

~Header: .*

The ~ negates the match.  .* is regex for "any character" (.), zero or 
more times (*).

There doesn't seem to be a way to setup a negative match regex directly 
from pan, thus the direct scorefile edit.  However, you could /start/ in 
pan, matching regex .*, and then just add the ~ to the entry it creates 
in the scorefile.

If that doesn't work, try .+ (+ is one or more times) instead of .*.

The problem however, is that pan runs the scoring algorithm on the 
overview, not on the entire post.  Technically, a news client does not 
fetch "headers", but rather "overviews", containing in general only those 
headers necessary to display and thread the listing in the overview 
(wrongly, header) pane, and to actually download the message if 
requested.  Headers not necessary for that, all the X- headers (which are 
considered unofficial) and headers such as nntp-posting-host, are not 
generally found in the overview (tho a server may optionally provide 
additional headers in the overview if they wish to).  Since pan doesn't 
rerun its scoring after actual message download, anything that's not 
found in the overview, doesn't count.  Thus, it's unfortunately 
impossible to score on them. 

This has been a pet peeve of mine since before pan even had scoring, when 
it was still (watch/normal/ignore) filters.  I'd have personally rather 
seen the ability to refilter after download added, since while one 
doesn't avoid the download then, one can still avoid actually reading the 
post if desired, than scoring, but I'm not the one writing the code, and 
Charles obviously had other priorities, so...  When I originally bugged 
it, the bug was set to target "bluesky", which means in effect, "It'd be 
nice, someday, but don't hold your breath.  Got a patch?"  Since I'm not 
a coder, other than a bit of bash, etc, I can't do the patch thing, which 
means I'm left waiting on someone else to find it equally irritating, and 
that's not happened yet, so... 

So keep it to the relatively narrow list of overview headers, and see if 
the ~Header: .* works.

Oh, well, pan's still the best option I've found, even without that 
feature, which I'd consider very useful.

Now, if gmane hasn't returned to 441s...

-- 
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