[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Procmail recipe for Nitwit unsubscribers who can't read DU sigs.

Incoming from Matthew Whitworth:
> s. keeling wrote:
> >:0 HB
> >* 1^0 ()(I will be out of the office|I will respond to your message when I 
> >return\.)
> >* 1^0 ^Subject:.*(un)?su(b)?(s)?cribe
> >* 1^0 $ ^^${SPCNL}*(un)?su(b)?scribe
> >* -1^0 ^Subject:.*Re:
> >{
> > LOG="(Un)?[twits] - "
> > :0
> > /dev/null
> >}
> Don't pipe to /dev/null just yet -- I've had five false positives since 
> implementing it this morning!
> I can provide procmail log entries (although not the actual emails -- I 
> /dev/null-ed prematurely) if you'd like.  I'm going to leave the filter 
> in place, but pipe them to a folder instead so that I can debug.

/dev/null is just the most satisfying place to consign them, not
necessarily the right place.  I assumed anyone wanting to use it would
think before they implemented it.  :-)

For instance, that "[twits]" in the LOG variable replaces what was
there before, else this message too would produce a false positive.
Then again, that may be wishful thinking and this may just be
hopelessly recursive and is doomed forever to fall into my spambin.

Other points:

  - we don't see messages on the list asking to be
    subscribed; they're not yet on the list so they can't post.

  - the variable SPCNL above is made up of:

    WSPC      = " 	"
    SPC       = "[$WSPC]"
    SPCNL     = "($SPC|$)"

    where WSPC is a space character plus a <TAB> character (in vi(m)?
    type CTRL-V<TAB>, and in emacs type CTRL-q<TAB>).

I don't know where I got the latter from originally, but Timo Salmi
would be a good place to start for things like this.

Any technology distinguishable from magic is insufficiently advanced.
(*)               http://www.spots.ab.ca/~keeling 
- -

Reply to: