Should archive server add Message-IDs to the cache?

Should archive server add Message-IDs to the cache?

[Hans de Graaff]

In one of the recent Gnus versions Lars changed caching of
Message-IDs. As a side effect, the archive server (which is nnfolder
based in my setup) now also adds Message-IDs to the ID cache.

This means that for all messages that I sent out, the Message-IDs are
put in the cache. If such a message comes back to me by some other
means (e.g. through a mailing list or an alias), then it will be
handled as a duplicate, even though I never saw the original, as I
don't read my archive server.

If I now add that I normaly 'delete duplicates, then my confusion
about 'disappearing' articles after upgrading becomes apparent. 

I have discussed this a bit with Lars, with my main end of the
argument being that the archive server should not add Message-IDs to
the cache of IDs. Lars suggested to bring it up on the list to see
what other people think.

Hans

Re: Should archive server add Message-IDs to the cache?

[Kai Grossjohann]

>>>>> Hans de Graaff writes:

  Hans> In one of the recent Gnus versions Lars changed caching of
  Hans> Message-IDs. As a side effect, the archive server (which is
  Hans> nnfolder based in my setup) now also adds Message-IDs to the
  Hans> ID cache.

I don't use archive servers, but I feel that the copy that's *not* in
the archive server should be considered the "original" and that the
copy in the archive server should be considered the "duplicate".  This
would mean that you won't be surprised about articles disappearing.

OTOH, though, deleting messages in the archive server isn't really
what you want, either, since then it wouldn't be an archive any
longer.  I'm not sure what setting nnmail-treat-duplicates to delete
means; does it mean marking as "read" or does it mean delete from the
face of the earth, as in "G DEL"?  Marking the articles as read
wouldn't be too bad, presumably.

Just my personal opinion,
kai
-- 
A large number of young women don't trust men with beards.
(BFBS Radio)

Re: Should archive server add Message-IDs to the cache?

[Sudish Joseph]

Kai Grossjohann writes:
> I don't use archive servers, but I feel that the copy that's *not* in
> the archive server should be considered the "original" and that the
> copy in the archive server should be considered the "duplicate".  This
> would mean that you won't be surprised about articles disappearing.

The archive server should have nothing to do with duplicate
detection.  Those messages are stored for a separate purpose, outside
of your normal message reading.  They aren't normal messages, at least
the way I use them.

> OTOH, though, deleting messages in the archive server isn't really
> what you want, either, since then it wouldn't be an archive any
> longer.

Nothing gets deleted from the archive server.  Quite the opposite,
it'll contain the only copy of your message in the gnus system (other
than the one in mail.duplicates).  

This breaks stuff like threading, "^", etc. in mailing list groups.  I
also find myself have to go and check the archive and duplicates
groups to see if the message even made it through.

Here's the picture.

rgnus                              gnus-5.4.22+ (and possibly earlier)
-----                              ------------
1 copy in archive                  1 copy in archive
1 copy in list group               1 copy in duplicates


Also, any funnction/backend/hook/whathaveyou that uses
nnml-request-accept-article from now on will add bogus entries into
the duplicate cache.


I guess the easy workaround for me is to turn off gnus' duplicate
detection and rely on external filters to weed out duplicates.

Eww,
-Sudish

Re: Should archive server add Message-IDs to the cache?

[David Moore]

Hans de Graaff <J.J.deGraaff@twi.tudelft.nl> writes:

> In one of the recent Gnus versions Lars changed caching of
> Message-IDs. As a side effect, the archive server (which is nnfolder
> based in my setup) now also adds Message-IDs to the ID cache.

> I have discussed this a bit with Lars, with my main end of the
> argument being that the archive server should not add Message-IDs to
> the cache of IDs. Lars suggested to bring it up on the list to see
> what other people think.

I don't think things in the Gcc line should be adding Message-IDs to the
cache.  Especially not for my real archive groups.  It makes duplicate
detection fairly useless for how I sometimes arrange things.

-- 
David Moore <dmoore@ucsd.edu>       | Computer Systems Lab      __o
UCSD Dept. Computer Science - 0114  | Work: (619) 534-8604    _ \<,_
La Jolla, CA 92093-0114             | Fax:  (619) 534-1445   (_)/ (_)
<URL:http://oj.egbt.org/dmoore/>    | In a cloud bones of steel.

Re: Should archive server add Message-IDs to the cache?

[Kai Grossjohann]

>>>>> David Moore writes:

  David> I don't think things in the Gcc line should be adding
  David> Message-IDs to the cache.  Especially not for my real archive
  David> groups.  It makes duplicate detection fairly useless for how
  David> I sometimes arrange things.

Only for those groups, I think.  I always Gcc into the group I post
from which makes duplicate deletion a nice thing to have.

What do you think?
kai
-- 
A large number of young women don't trust men with beards.
(BFBS Radio)

Re: Should archive server add Message-IDs to the cache?

[David Moore]

Kai Grossjohann <grossjohann@charly.informatik.uni-dortmund.de> writes:

> >>>>> David Moore writes:
> 
>   David> I don't think things in the Gcc line should be adding
>   David> Message-IDs to the cache.  Especially not for my real archive
>   David> groups.  It makes duplicate detection fairly useless for how
>   David> I sometimes arrange things.
> 
> Only for those groups, I think.  I always Gcc into the group I post
> from which makes duplicate deletion a nice thing to have.

No, it's global.  And it has to be, since it has to know about the
duplication before it's decided what group the message belongs in, since
you want to be able to use the `Gnus-Warning: duplicate message...'
header when filing your mail.

-- 
David Moore <dmoore@ucsd.edu>       | Computer Systems Lab      __o
UCSD Dept. Computer Science - 0114  | Work: (619) 534-8604    _ \<,_
La Jolla, CA 92093-0114             | Fax:  (619) 534-1445   (_)/ (_)
<URL:http://oj.egbt.org/dmoore/>    | In a cloud bones of steel.

Re: Should archive server add Message-IDs to the cache?

[Kai Grossjohann]

>>>>> David Moore writes:

  David> I don't think things in the Gcc line should be adding
  David> Message-IDs to the cache.  Especially not for my real archive
  David> groups.  It makes duplicate detection fairly useless for how
  David> I sometimes arrange things.

>>>>> Kai Grossjohann <grossjohann@charly.informatik.uni-dortmund.de> writes:

  Kai> Only for those groups, I think.  I always Gcc into the group I
  Kai> post from which makes duplicate deletion a nice thing to have.

  David> No, it's global.  And it has to be, since it has to know
  David> about the duplication before it's decided what group the
  David> message belongs in, since you want to be able to use the
  David> `Gnus-Warning: duplicate message...'  header when filing your
  David> mail.

So there seems to be a conflict.  Some people use archive groups, like
you, and would like one behavior.  Other people don't use archive
groups (like me) and would like the other behavior.

But I guess this could be made an option, couldn't it?

kai
-- 
A large number of young women don't trust men with beards.
(BFBS Radio)

Re: Should archive server add Message-IDs to the cache?

[Paul Franklin]

>>>>> Kai Grossjohann writes:

 > So there seems to be a conflict.  Some people use archive groups, like
 > you, and would like one behavior.  Other people don't use archive
 > groups (like me) and would like the other behavior.

 > But I guess this could be made an option, couldn't it?

I'd like to see this made an option as well.  Right now, I always bcc:
myself to make sure my outgoing and incoming mail gets merged (which
is almost the same as what Kai wants).  I've just never been brave
enough to convert over to using gcc:, and having this option become
"official" would probably motivate me to make that transition.

--Paul

Re: Should archive server add Message-IDs to the cache?

[Lars Magne Ingebrigtsen]

Kai Grossjohann <grossjohann@ls6.informatik.uni-dortmund.de> writes:

> So there seems to be a conflict.  Some people use archive groups, like
> you, and would like one behavior.  Other people don't use archive
> groups (like me) and would like the other behavior.

I've switched the caching of Message-IDs off on Gcc.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@ifi.uio.no * Lars Ingebrigtsen

Re: Should archive server add Message-IDs to the cache?

[Fredrik Glockner]

[Lars]
|
| I've switched the caching of Message-IDs off on Gcc.

Could this be an option, please?  I did indeed like this behavior.

Fredrik

Re: Should archive server add Message-IDs to the cache?

[Lars Magne Ingebrigtsen]

fredrigl@math.uio.no (Fredrik Glockner) writes:

> | I've switched the caching of Message-IDs off on Gcc.
> 
> Could this be an option, please?  I did indeed like this behavior.

Ok, I've now added a new variable to control this --
`nnmail-cache-accepted-message-ids'.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@ifi.uio.no * Lars Ingebrigtsen