Diagnosis of a nnmail-split-methods problem

Diagnosis of a nnmail-split-methods problem

[Joseph Barillari]

Hi.

I'm curious if anyone on the list can provide some tips on diagnosing
a nnmail-split-methods problem.

My nnmail-split-methods variable looks like this:


(setq nnmail-split-methods
      '(
	("mail.spam" "^X-Rocket-Spam: .*")
	("mail.spam" "^X-Spam-Status: Yes.*")
	("mail.virus" "^X-JDB-Virus-Warning:.*")
	("mail.spam" "^X-YahooFilteredBulk: .*")
	("mail.receipts" "^Subject: Return Receipt.*")

        ;; <snip loads of other classifiers>

	("mail.mail" "")
	)
      )

The ("mail.virus" "^X-JDB-Virus-Warning:.*") entry is new. I added it
today to filter out the dozens of SoBig viruses that have been
crashing into my inbox. (There's a procmail rule that adds it to the
messages.)

While the other classifiers typically work, this new 'virus'
classifier does not. Sometimes, messages with the X-JDB-Virus-Warning
header get dumped into mail.virus, as they should be. Other times,
they miss that filter and make it into mail.mail.

The /really/ strange part is that if I hit 'B r' to respool the
misclassified messages in mail.mail, gnus correctly puts them in the
virus folder.

Any hints as to how to diagnose this would be appreciated. I'm running
the latest gnus from CVS.

best,

--Joe

-- 
Joseph Barillari -- http://barillari.org

Re: Diagnosis of a nnmail-split-methods problem

[Kai Großjohann]

Joseph Barillari <jbarilla@princeton.edu> writes:

> (setq nnmail-split-methods
>       '(
> 	("mail.spam" "^X-Rocket-Spam: .*")
> 	("mail.spam" "^X-Spam-Status: Yes.*")
> 	("mail.virus" "^X-JDB-Virus-Warning:.*")
> 	("mail.spam" "^X-YahooFilteredBulk: .*")
> 	("mail.receipts" "^Subject: Return Receipt.*")
>
>         ;; <snip loads of other classifiers>
>
> 	("mail.mail" "")
> 	)
>       )

One difference between the virus rule and the other rules is that the
virus rule does NOT have a space (or tab?) after the colon.

I would be surprised, but you never know...

Btw, I think the ".*" at the end are not needed.  (".*" at the end
might be useful for fancy splitting.)  The string "xyzzy" matches the
regexp "x" just like it matches the regexp "x.*".
-- 
Two cafe au lait please, but without milk.

Re: Diagnosis of a nnmail-split-methods problem

[Joseph Barillari]

>>>>> "KG" == Kai Großjohann <kai.grossjohann@gmx.net> writes:

    KG> One difference between the virus rule and the other rules is
    KG> that the virus rule does NOT have a space (or tab?) after the
    KG> colon.

    KG> I would be surprised, but you never know...

I can insert a space and see what happens. The flood of viruses has
subsided, so I may not have much testing material. But the odd thing
about the problem was its intermittentcy. Sometimes it would split the
messages properly; other times, on messages with an identical
virus-flag header, it would dump them in the regular mail folder.

    KG> Btw, I think the ".*" at the end are not needed.  (".*" at the
    KG> end might be useful for fancy splitting.)  The string "xyzzy"
    KG> matches the regexp "x" just like it matches the regexp "x.*".

Gotcha.

Thanks,

--Joe


-- 
Joseph Barillari -- http://barillari.org