Fallback for nnmail-split-methods

Fallback for nnmail-split-methods

[Thomas Gehrlein]

nnmail-split-fancy has a fallback option for mails that don't match any of the
splitting rules.  nnmail-split-methods doesnt't provide a fallback.  I
discovered this after some of my mails had disappeared.

Here's some code that makes nnmail-split-methods safer.  I'm not really happy
with my approach, but I couldn't figure out how the gnus splitting code works.

(defun nnmail-split-methods-add-fall-back ()
  "Add a fall-back group to nnmail-split-methods if there isn't one."
  ;; check for fall-back: second element of last list of split methods is "".
  (unless (or (eq nnmail-split-methods 'nnmail-split-fancy)
              (string= ""
                       (cadr (elt nnmail-split-methods
                                  (1- (length nnmail-split-methods))))))
    ;; add fall-back if there isn't one
    (setq nnmail-split-methods
          (append nnmail-split-methods (list (list "mail.misc" ""))))))

(add-hook 'nnmail-prepare-incoming-hook 'nnmail-split-methods-add-fall-back)

Thomas

Re: Fallback for nnmail-split-methods

[Kai Großjohann]

Thomas.Gehrlein@t-online.de (Thomas Gehrlein) writes:

> nnmail-split-fancy has a fallback option for mails that don't match any of the
> splitting rules.  nnmail-split-methods doesnt't provide a fallback.  I
> discovered this after some of my mails had disappeared.

It appears that there is now some code which is supposed to do this:
the last group mentioned in nnmail-split-methods is used as the
fallback.

Can someone confirm this?

There is the following comment in the code:

	  ;; Go through the split methods to find a match.

Some lines down, there is this snippet:

	      ;; This is the final group, which is used as a
	      ;; catch-all.
	      (unless group-art
		(setq group-art
		      (list (cons (car method)
				  (funcall func (car method)))))))))

I think this does the trick, but I haven't tested it.  Could you test
it?

kai
-- 
Silence is foo!

Re: Fallback for nnmail-split-methods

[Frank Haun]

Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) writes:

> Thomas.Gehrlein@t-online.de (Thomas Gehrlein) writes:
> 
> > nnmail-split-fancy has a fallback option for mails that don't match
> > any of the splitting rules. nnmail-split-methods doesnt't provide a
> > fallback. I discovered this after some of my mails had disappeared.
> 
> It appears that there is now some code which is supposed to do this:
> the last group mentioned in nnmail-split-methods is used as the
> fallback.
> 
> Can someone confirm this?
> 
> There is the following comment in the code:
> 
> 	  ;; Go through the split methods to find a match.
> 
> Some lines down, there is this snippet:
> 
> 	      ;; This is the final group, which is used as a
> 	      ;; catch-all.
> 	      (unless group-art
> 		(setq group-art
> 		      (list (cons (car method)
> 				  (funcall func (car method)))))))))
> 
> I think this does the trick, but I haven't tested it.  Could you test
> it?

I've tested it, and it doesn't work correctly.

For testing I've made some cron mails:

| From: fhaun (Cron Daemon)
| To: fhaun
| Subject: Cron <fhaun@pille> echo "Hello Frank"

with this split settings:

(setq nnmail-split-methods
      '(
	("CronJobs" "^\\(From:\\).*Cron Daemon")
	("MailingList-cdwrite" "^\\(To:\\|Cc:\\).*cdwrite.*debian.org")
	("MailingList-cygwin-announce" "^\\(To:\\|From:\\).*cygwin-announce")
	("SYSTEM" "^\\(From:\\).*Charlie Root <root>")
	("Unsorted" "")))

the cron mails go to "CronJobs", good. Then I set the split to:

(setq nnmail-split-methods
      '(
	("SYSTEM" "^\\(From:\\).*Charlie Root <root>")))

and the cron mails go to "SYSTEM", good. Then I set the split to:

(setq nnmail-split-methods
      '(
	("MailingList-cdwrite" "^\\(To:\\|Cc:\\).*cdwrite.*debian.org")
	("MailingList-cygwin-announce" "^\\(To:\\|From:\\).*cygwin-announce")
	("SYSTEM" "^\\(From:\\).*Charlie Root <root>")

and all the cron mails vanished, bad.

Message Buffer says:
  nnml: Reading incoming mail from file...
  nnml: Reading incoming mail (1 new)...done

Frank
-- 
  line-move: End of buffer [42 times]

Re: Fallback for nnmail-split-methods

[Kai Großjohann]

Frank Haun <pille3000@gmx.net> writes:

> (setq nnmail-split-methods
>       '(
> 	("SYSTEM" "^\\(From:\\).*Charlie Root <root>")))
>
> and the cron mails go to "SYSTEM", good. Then I set the split to:
>
> (setq nnmail-split-methods
>       '(
> 	("MailingList-cdwrite" "^\\(To:\\|Cc:\\).*cdwrite.*debian.org")
> 	("MailingList-cygwin-announce" "^\\(To:\\|From:\\).*cygwin-announce")
> 	("SYSTEM" "^\\(From:\\).*Charlie Root <root>")
>
> and all the cron mails vanished, bad.

I think this means that the fallback mechanism works if
nnmail-split-methods has only one element, whereas it fails if it has
more than one element.

Hm.

kai
-- 
Silence is foo!

Re: Fallback for nnmail-split-methods

[Paul Jarc]

Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) wrote:
> Thomas.Gehrlein@t-online.de (Thomas Gehrlein) writes:
>> nnmail-split-fancy has a fallback option for mails that don't match
>> any of the splitting rules.  nnmail-split-methods doesnt't provide
>> a fallback.  I discovered this after some of my mails had
>> disappeared.
>
> It appears that there is now some code which is supposed to do this:
> the last group mentioned in nnmail-split-methods is used as the
> fallback.

Wouldn't it be less surprising and more reliable to use a constant
fallback "bogus", just like nnmail-split-fancy?


paul

Re: Fallback for nnmail-split-methods

[Kai Großjohann]

prj@po.cwru.edu (Paul Jarc) writes:

> Wouldn't it be less surprising and more reliable to use a constant
> fallback "bogus", just like nnmail-split-fancy?

Good idea, actually.  Especially since it seems that the fallback
only works when (= (length nnmail-split-methods) 1).

kai
-- 
Silence is foo!