Imap split, does not work (of course)

Imap split, does not work (of course)

[Francis Moreau]

Hello,

I was struggling during one hour to setup split stuff in my IMAP mail
box, and as usually I'm wasting my time since I'm, well..., not very
lucky with gnus.

So here's my config:

(setq gnus-secondary-select-methods
      '((nnimap "gmail"
	(nnimap-inbox "nnimap+gmail:INBOX")
	(nnimap-address "imap.gmail.com")
	(nnimap-server-port 993)
	(nnimap-stream ssl)
	(nnimap-authinfo-file "~/.authinfo")
	(nnir-search-engine imap)
	(nnimap-split-methods
	 (("INBOX.toto" ""))))))

I also tried different variants like:

  (nnimap-inbox "INBOX")

but all failed.

Could anybody tell me what I am again missing ?

Thanks
-- 
Francis

Re: Imap split, does not work (of course)

[Richard Riley]

Francis Moreau <francis.moro@gmail.com> writes:

> Hello,
>
> I was struggling during one hour to setup split stuff in my IMAP mail
> box, and as usually I'm wasting my time since I'm, well..., not very
> lucky with gnus.
>
> So here's my config:
>
> (setq gnus-secondary-select-methods
>       '((nnimap "gmail"
> 	(nnimap-inbox "nnimap+gmail:INBOX")
> 	(nnimap-address "imap.gmail.com")
> 	(nnimap-server-port 993)
> 	(nnimap-stream ssl)
> 	(nnimap-authinfo-file "~/.authinfo")
> 	(nnir-search-engine imap)
> 	(nnimap-split-methods
> 	 (("INBOX.toto" ""))))))
>
> I also tried different variants like:
>
>   (nnimap-inbox "INBOX")
>
> but all failed.
>
> Could anybody tell me what I am again missing ?
>

I too would love to find out. I have an open question about something
called "backend slots" which is related to this.

,----
|      ;; (setq nnimap-split-inbox '("INBOX"))
|      ;; (setq nnimap-split-fancy '(| (: spam-split) "INBOX"))
|      ;; (setq nnimap-split-methods 'nnimap-split-fancy)
|      ;; (setq nnimap-split-download-body t)
|     
|      ;; (setq nnmail-split-inbox '("INBOX"))
|      ;; (setq nnmail-split-fancy '(| (: spam-split) "SYS-INBOX"))
|      ;; (setq nnmail-split-methods 'nnmail-split-fancy)
|      
|      ;; (setq spam-split-disabled nil)
|      
|      ;; (setq spam-autodetect-recheck-messages t)
|      ;; (setq spam-mark-ham-unread-before-move-from-spam-group t)
|      ;; (setq spam-mark-new-messages-in-spam-group-as-spam t)
|      
`----

Also its not entirely clear to me if this complements the spam
processing settings set in group customisation using G c.

Re: Imap split, does not work (of course)

[Lars Ingebrigtsen]

Francis Moreau <francis.moro@gmail.com> writes:

> (setq gnus-secondary-select-methods
>       '((nnimap "gmail"
> 	(nnimap-inbox "nnimap+gmail:INBOX")
> 	(nnimap-address "imap.gmail.com")
> 	(nnimap-server-port 993)
> 	(nnimap-stream ssl)
> 	(nnimap-authinfo-file "~/.authinfo")
> 	(nnir-search-engine imap)
> 	(nnimap-split-methods
> 	 (("INBOX.toto" ""))))))

Let's indent and edit out all the obviously superfluous and unused slots
first:

(setq gnus-secondary-select-methods
      '((nnimap "gmail"
                 (nnimap-inbox "nnimap+gmail:INBOX")
                 (nnimap-address "imap.gmail.com")
                 (nnimap-split-methods
                  (("INBOX.toto" ""))))))

Ok, that should be sufficient.

Then let's fix it up:

(setq gnus-secondary-select-methods
      '((nnimap "gmail"
                 (nnimap-inbox "INBOX")
                 (nnimap-address "imap.gmail.com")
                 (nnimap-split-methods
                  (("INBOX.toto" ""))))))

But now I have to ask you: What did you mean to express by saying:

(nnimap-split-methods
 (("INBOX.toto" "")))

I'm really curious. 

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Poll time: IMAP client-side default match (was: Imap split, does not work (of course))

[Lars Ingebrigtsen]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> But now I have to ask you: What did you mean to express by saying:
>
> (nnimap-split-methods
>  (("INBOX.toto" "")))

Oh, you meant that you wanted all the mail to be moved to INBOX.toto?
That's certainly reasonable, and, as you've found out, doesn't work.

And it's not documented why it doesn't work, because it not working is
experimental, and I forgot about it.

The issue is this: When using non-IMAP splitting, Gnus really needs to
have a "default split", or a catch-all rule at the end.  If it doesn't
exist, Gnus makes one.  Otherwise Gnus would have nowhere to put the
mail it pulls down from, say, POP3, and which matches no (other) rules,
and the mail would be lost.

On IMAP, on the other hand, Gnus can just leave the mail in INBOX, and
no harm would be done.  (And I think that's what IMAP users expect.)

So the problem is: If you have a split that you're using both for POP3
and IMAP, then POP3 one needs a default split, while the IMAP shouldn't
have one.

Gnus deals with this by special-casing the IMAP case.  It simply ignores
"" (catch-all) rules.

We could deal with this by:

1) Just document that IMAP client-side splits ignores the catch-all
rule.

2) Something else.  Ideas are welcome.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: Poll time: IMAP client-side default match

[Rupert Swarbrick]

[-- Attachment #1: Type: text/plain, Size: 658 bytes --]

Lars Ingebrigtsen <larsi@gnus.org> writes:
> So the problem is: If you have a split that you're using both for POP3
> and IMAP, then POP3 one needs a default split, while the IMAP shouldn't
> have one.
>
> Gnus deals with this by special-casing the IMAP case.  It simply ignores
> "" (catch-all) rules.
>
> We could deal with this by:
>
> 1) Just document that IMAP client-side splits ignores the catch-all
> rule.
>
> 2) Something else.  Ideas are welcome.

Couldn't you say that catch-all rules are followed if they exist on
IMAP? (And if there isn't one, stuff is left in INBOX, as you described)
I don't understand why this doesn't make sense...

Rupert

[-- Attachment #2: Type: application/pgp-signature, Size: 315 bytes --]

Re: Poll time: IMAP client-side default match

[Lars Ingebrigtsen]

Rupert Swarbrick <rswarbrick@gmail.com> writes:

> Couldn't you say that catch-all rules are followed if they exist on
> IMAP? (And if there isn't one, stuff is left in INBOX, as you described)
> I don't understand why this doesn't make sense...

I use the same split methods on nnml and nnimap.  One needs defaults,
and one probably shouldn't have one.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: Imap split, does not work (of course)

[Francis Moreau]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> Francis Moreau <francis.moro@gmail.com> writes:
>
>> (setq gnus-secondary-select-methods
>>       '((nnimap "gmail"
>> 	(nnimap-inbox "nnimap+gmail:INBOX")
>> 	(nnimap-address "imap.gmail.com")
>> 	(nnimap-server-port 993)
>> 	(nnimap-stream ssl)
>> 	(nnimap-authinfo-file "~/.authinfo")
>> 	(nnir-search-engine imap)
>> 	(nnimap-split-methods
>> 	 (("INBOX.toto" ""))))))
>
> Let's indent and edit out all the obviously superfluous and unused slots
> first:
>
> (setq gnus-secondary-select-methods
>       '((nnimap "gmail"
>                  (nnimap-inbox "nnimap+gmail:INBOX")
>                  (nnimap-address "imap.gmail.com")
>                  (nnimap-split-methods
>                   (("INBOX.toto" ""))))))
>
> Ok, that should be sufficient.
>
> Then let's fix it up:
>
> (setq gnus-secondary-select-methods
>       '((nnimap "gmail"
>                  (nnimap-inbox "INBOX")
>                  (nnimap-address "imap.gmail.com")
>                  (nnimap-split-methods
>                   (("INBOX.toto" ""))))))

Unfortunately that doesn't work, I already tried this yesterday during
my 'battle' and I did retest this today with no success.

BTW before retesting, I restored my previous value of
nnimap-split-methods:

    (nnimap-split-methods
        (("INBOX.toto" "^From:.*francis.moro@gmail.com")))

so I want Gnus to move all mails from me to go to the 'toto' folder.

>
> But now I have to ask you: What did you mean to express by saying:
>
> (nnimap-split-methods
>  (("INBOX.toto" "")))
>
> I'm really curious. 

Well, since I had and still have no idea why Gnus refuses to split my
emails, I tried to simplify my configuration. Hence I tried to match all
received emails just to be sure that the issue wasn't coming from the
regexp.

Thanks
-- 
Francis

Re: Imap split, does not work (of course)

[Richard Riley]

Francis Moreau <francis.moro@gmail.com> writes:

> Lars Ingebrigtsen <larsi@gnus.org> writes:
>
>> Francis Moreau <francis.moro@gmail.com> writes:
>>
>>> (setq gnus-secondary-select-methods
>>>       '((nnimap "gmail"
>>> 	(nnimap-inbox "nnimap+gmail:INBOX")
>>> 	(nnimap-address "imap.gmail.com")
>>> 	(nnimap-server-port 993)
>>> 	(nnimap-stream ssl)
>>> 	(nnimap-authinfo-file "~/.authinfo")
>>> 	(nnir-search-engine imap)
>>> 	(nnimap-split-methods
>>> 	 (("INBOX.toto" ""))))))
>>
>> Let's indent and edit out all the obviously superfluous and unused slots
>> first:
>>
>> (setq gnus-secondary-select-methods
>>       '((nnimap "gmail"
>>                  (nnimap-inbox "nnimap+gmail:INBOX")
>>                  (nnimap-address "imap.gmail.com")
>>                  (nnimap-split-methods
>>                   (("INBOX.toto" ""))))))
>>
>> Ok, that should be sufficient.
>>
>> Then let's fix it up:
>>
>> (setq gnus-secondary-select-methods
>>       '((nnimap "gmail"
>>                  (nnimap-inbox "INBOX")
>>                  (nnimap-address "imap.gmail.com")
>>                  (nnimap-split-methods
>>                   (("INBOX.toto" ""))))))
>
> Unfortunately that doesn't work, I already tried this yesterday during
> my 'battle' and I did retest this today with no success.
>
> BTW before retesting, I restored my previous value of
> nnimap-split-methods:
>
>     (nnimap-split-methods
>         (("INBOX.toto" "^From:.*francis.moro@gmail.com")))
>
> so I want Gnus to move all mails from me to go to the 'toto' folder.
>
>>
>> But now I have to ask you: What did you mean to express by saying:
>>
>> (nnimap-split-methods
>>  (("INBOX.toto" "")))
>>
>> I'm really curious. 
>
> Well, since I had and still have no idea why Gnus refuses to split my
> emails, I tried to simplify my configuration. Hence I tried to match all
> received emails just to be sure that the issue wasn't coming from the
> regexp.
>

Hi Francis,

Do you have any residual settings set in the "G c" (configure group from
the Group buffer) configuration options that might be causing a clash?

regards

r.

Re: Imap split, does not work (of course)

[Francis Moreau]

Hello Richard,

Richard Riley <rileyrg@googlemail.com> writes:

> Francis Moreau <francis.moro@gmail.com> writes:

[...]

>> Well, since I had and still have no idea why Gnus refuses to split my
>> emails, I tried to simplify my configuration. Hence I tried to match all
>> received emails just to be sure that the issue wasn't coming from the
>> regexp.
>>
>
> Do you have any residual settings set in the "G c" (configure group from
> the Group buffer) configuration options that might be causing a clash?

Unfortunately not.

Thanks
-- 
Francis

Re: Imap split, does not work (of course)

[Francis Moreau]

Hi Richard,

Richard Riley <rileyrg@googlemail.com> writes:

[...]

>
> Do you have any residual settings set in the "G c" (configure group from
> the Group buffer) configuration options that might be causing a clash?

I eventually found something.

Beside my nnimap split config, I had this in my .gnus set after my IMAP
split config:

   (setq nnmail-split-lowercase-expanded t)
   (setq nnmail-crosspost t)
   (setq nnmail-split-methods
         '(("save.wyplay.bugs" "^From: bugzilla-daemon@foo.com")
           ("Inbox" "")))

I commented out this part, and splitting in my IMAP mail box magically
works...

I now have set this in my .bashrc:

   alias gnus=gandalf

-- 
Francis

Re: Imap split, does not work (of course)

[Francis Moreau]

Francis Moreau <francis.moro@gmail.com> writes:

> Lars Ingebrigtsen <larsi@gnus.org> writes:
>

[...]

>
> Unfortunately that doesn't work, I already tried this yesterday during
> my 'battle' and I did retest this today with no success.
>
> BTW before retesting, I restored my previous value of
> nnimap-split-methods:
>
>     (nnimap-split-methods
>         (("INBOX.toto" "^From:.*francis.moro@gmail.com")))
>
> so I want Gnus to move all mails from me to go to the 'toto' folder.
>
>>
>> But now I have to ask you: What did you mean to express by saying:
>>
>> (nnimap-split-methods
>>  (("INBOX.toto" "")))
>>
>> I'm really curious. 
>
> Well, since I had and still have no idea why Gnus refuses to split my
> emails, I tried to simplify my configuration. Hence I tried to match all
> received emails just to be sure that the issue wasn't coming from the
> regexp.

Could anybody point out some way to debug this ?

Any traces to activate to show some revelant stuff ?

Thanks
-- 
Francis

Re: Imap split, does not work (of course)

[Eric S Fraga]

Francis Moreau <francis.moro@gmail.com> writes:

[...]

>>     (nnimap-split-methods
>>         (("INBOX.toto" "^From:.*francis.moro@gmail.com")))

[...]

> Could anybody point out some way to debug this ?
>
> Any traces to activate to show some revelant stuff ?
>
> Thanks

I cannot help you wrt debugging; I can however show you my (abridged)
configuration which is based on fancy splitting.  I never did get
"not-so-fancy" splitting to work, so I can understand your frustration.

All split related variables in my configuration are here:

--8<---------------cut here---------------start------------->8---
(setq
  nnmail-split-fancy 
      '(| (to ".*somelist@somehost.org" "lists")
	  (from ".*someuser@some.other.host" "people")
	 "general"
	 )
  nnimap-split-methods 'default
  nnmail-split-methods 'nnmail-split-fancy
  )
--8<---------------cut here---------------end--------------->8---

this works just fine for me.  I have nothing specified in the select
methods variables and nothing in the group configurations either, as far
as I am aware.

HTH,
Eric

-- 
: Eric S Fraga (GnuPG: 0xC89193D8FFFCF67D) in Emacs 24.0.50.1 + No Gnus v0.11

Re: Imap split, does not work (of course)

[Richard Riley]

Eric S Fraga <e.fraga@ucl.ac.uk> writes:

> Francis Moreau <francis.moro@gmail.com> writes:
>
> [...]
>
>>>     (nnimap-split-methods
>>>         (("INBOX.toto" "^From:.*francis.moro@gmail.com")))
>
> [...]
>
>> Could anybody point out some way to debug this ?
>>
>> Any traces to activate to show some revelant stuff ?
>>
>> Thanks
>
> I cannot help you wrt debugging; I can however show you my (abridged)
> configuration which is based on fancy splitting.  I never did get
> "not-so-fancy" splitting to work, so I can understand your frustration.
>
> All split related variables in my configuration are here:
>
>
>
> --8<---------------cut here---------------start------------->8---
> (setq
>   nnmail-split-fancy 
>       '(| (to ".*somelist@somehost.org" "lists")
> 	  (from ".*someuser@some.other.host" "people")
> 	 "general"
> 	 )
>   nnimap-split-methods 'default
>   nnmail-split-methods 'nnmail-split-fancy
>   )
> --8<---------------cut here---------------end--------------->8---
>
>
>

If someone could explain the concept of "backend slots" vs using "setq"
it would help me a lot.

I have been unable to see when something should be set with setq and
when things are passed as a parameter to secondary-methods call.

Re: Imap split, does not work (of course)

[Michael Welsh Duggan]

Richard Riley <rileyrg@googlemail.com> writes:

> If someone could explain the concept of "backend slots" vs using "setq"
> it would help me a lot.

Well, in the manual they are not called "backend slots".  They are
called "server variables", or "server parameters".  There is even an
info page devoted to describing this, section 6.1.5.

Their purpose is as variables that can differ from server to server if
your setup contains multiple servers.  These variables are set as part
of the `gnus-select-method' or `gnus-secondary-select-method's.  A
select method is a list consisting of a symbol, a name, and then an
optional set of "server variable" pairs.  An example from the docs:

     (nnml "public"
           (nnml-directory "~/my-mail/")
           (nnml-active-file "~/my-mail/active")
           (nnml-newsgroups-file "~/my-mail/newsgroups"))

The `nnml' symbol is the actual select method.  The name of the server
is "public", and it set three server variables associated with this
server.  

> I have been unable to see when something should be set with setq and
> when things are passed as a parameter to secondary-methods call.

Now, the `gnus-select-method' is usually set using `setq'.  For the
above example:

(setq gnus-select-method 
      '(nnml "public"
             (nnml-directory "~/my-mail/")
             (nnml-active-file "~/my-mail/active")
             (nnml-newsgroups-file "~/my-mail/newsgroups")))

On the other hand, `gnus-secondary-select-method' is a set of servers,
so it would set there like this:

(setq gnus-select-method 
      '((nnml "public"
              (nnml-directory "~/my-mail/")
              (nnml-active-file "~/my-mail/active")
              (nnml-newsgroups-file "~/my-mail/newsgroups"))))

Note the extra pair of parentheses.  A second would be added like this:

(setq gnus-select-method 
      '((nnml "public"
              (nnml-directory "~/my-mail/")
              (nnml-active-file "~/my-mail/active")
              (nnml-newsgroups-file "~/my-mail/newsgroups"))
        (nnbogus "bogus"
                 (bogus-var-1 "string value")
                 (bogus-var-1 symbol-value))))


All this probably went into more detail than you actually needed, but I
feel it is better to give too much information than too little.  (I now
wait for someone to tell me that I am mistaken about what was meant by
"backend slots" in the first place.)

-- 
Michael Welsh Duggan
(md5i@md5i.com)

Re: Imap split, does not work (of course)

[Steinar Bang]

> Richard Riley <rileyrg@googlemail.com> writes:

>> If someone could explain the concept of "backend slots" vs using "setq"
>> it would help me a lot.

Slots are fields in Common Lisp-style "structs"
 http://www.gnu.org/software/emacs/manual/html_node/cl/Structures.html

Whether that's the case for the gnus backend slots as well, I don't
know.  But it seems likely.

Re: Imap split, does not work (of course)

[Richard Riley]

Steinar Bang <sb@dod.no> writes:

>> Richard Riley <rileyrg@googlemail.com> writes:
>
>>> If someone could explain the concept of "backend slots" vs using "setq"
>>> it would help me a lot.
>
> Slots are fields in Common Lisp-style "structs"
>  http://www.gnu.org/software/emacs/manual/html_node/cl/Structures.html
>
> Whether that's the case for the gnus backend slots as well, I don't
> know.  But it seems likely.

Thanks to you and Michael. Unfortunately, It doesn't really clarify
things too much (to me) since the examples given here and elsewhere seem
to alternate between using setq and these slots. It's not apparent to me
when slots should be used : I had always (I guess incorrectly) assumed
from other examples that if I use a "global" like

(setq nnmail-xyz "blah")

Then the "back end slot" would cleverly default to this if I didn't
specifically set it in select-method parameters. Indeed, I posted my
setup numerous times and had it commented on by some pretty knowledgable
Gnus users before the recent thing about it being wrong since some
are/should be back-end slots.

I think I'll start my set up from scratch. It all used to work fine so
I'm not exactly sure which part is broken.

Cheers
r.