Request: ding subscribers' group parameter

Request: ding subscribers' group parameter

[Karl Kleinpaste]

Could folks who subscribe here prettyplease put this parameter in
the group you use for the list?

(to-address . "ding@gnus.org")

It will prevent generation of private duplicates when `f'ollowing up
to the list.

*Info* -> Gnus -> Group Buffer -> Group Parameters (to-address)

thanx,
--karl

Re: Request: ding subscribers' group parameter

[Paul Jarc]

Karl Kleinpaste <karl@charcoal.com> writes:
> Could folks who subscribe here prettyplease put this parameter in
> the group you use for the list?
> 
> (to-address . "ding@gnus.org")

Since Gnus now respects Mail-Followup-To, it'd be better to encourage
people to add that field to outgoing messages.  And, of course, it'd
be even better if Gnus would generate that field automatically.

MFT is supposed to be generated after editing the message, just before
sending, and it should contain all the To+Cc addresses (but if the
user provided one maually, we leave it alone).  I think the best way
to do this would be to look at the current group's parameters; if
there is a to-address or to-list parameter and its address is included
in To+Cc, then we should generate MFT.  It seems this would have to be
done in message-send or message-send-mail, possibly with help from
message-get-reply-headers... I think.  But it doesn't quite seem
appropriate for message.el to be looking at Gnus group parameters.
I'm willing to write the code if we can get these issues sorted out.
Ideas?


paul

Re: Request: ding subscribers' group parameter

[Colin Walters]

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

> Since Gnus now respects Mail-Followup-To, it'd be better to
> encourage people to add that field to outgoing messages.  And, of
> course, it'd be even better if Gnus would generate that field
> automatically.

Here's what I do in my `gnus-posting-styles':

	((and (message-mail-p) (stringp gnus-newsgroup-name))
	 ("Mail-Followup-To" (gnus-group-get-parameter gnus-newsgroup-name 'to-address)))

Re: Request: ding subscribers' group parameter

[Paul Jarc]

Colin Walters <walters@cis.ohio-state.edu> writes:
> Here's what I do in my `gnus-posting-styles':
> 
> 	((and (message-mail-p) (stringp gnus-newsgroup-name))
> 	 ("Mail-Followup-To" (gnus-group-get-parameter gnus-newsgroup-name 'to-address)))

Neat, but not what I was looking for.  Mail-Followup-To should
normally be preserved from the previous message.  It need not be only
a single address; someone may actually want an extra copy of a thread.
Since To+Cc is initialized from the previous MFT, the new MFT should
be generated from To+Cc.


paul

Re: Request: ding subscribers' group parameter

[Kai Großjohann]

On Tue, 31 Jul 2001, Paul Jarc wrote:

> MFT is supposed to be generated after editing the message, just
> before sending, and it should contain all the To+Cc addresses (but
> if the user provided one maually, we leave it alone).  I think the
> best way to do this would be to look at the current group's
> parameters; if there is a to-address or to-list parameter and its
> address is included in To+Cc, then we should generate MFT.  It seems
> this would have to be done in message-send or message-send-mail,
> possibly with help from message-get-reply-headers... I think.  But
> it doesn't quite seem appropriate for message.el to be looking at
> Gnus group parameters.  I'm willing to write the code if we can get
> these issues sorted out.  Ideas?

Gcc handling looks at the parameters, too.  So you might want to look
how that is done?

I guess it all amounts to adding stuff to message-send-hook or
message-send-actions.

I have to admit that I'm a little surprised about this explanation.
Before I knew anything about MFT (except its name), I thought the
semantics would allow for saying something like `don't cc me!', rather
than inversely specifying whom to send mail to.  Ah, I get older than
a cow, and still learn something, now.¹

kai

¹ Crippled translation from a German saying, but, hey, it rhymes.
-- 
~/.signature: No such file or directory

Re: Request: ding subscribers' group parameter

[Lars Magne Ingebrigtsen]

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

> I have to admit that I'm a little surprised about this explanation.
> Before I knew anything about MFT (except its name), I thought the
> semantics would allow for saying something like `don't cc me!', rather
> than inversely specifying whom to send mail to.  

Well, isn't that what Mail-Copies-To does?  As opposed to
Mail-Followup-To? 

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

Re: Request: ding subscribers' group parameter

[Kai Großjohann]

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

> Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) writes:
>
>> I have to admit that I'm a little surprised about this explanation.
>> Before I knew anything about MFT (except its name), I thought the
>> semantics would allow for saying something like `don't cc me!', rather
>> than inversely specifying whom to send mail to.  
>
> Well, isn't that what Mail-Copies-To does?  As opposed to
> Mail-Followup-To? 

I don't know.  I thought that MCT was for news only, and using it in
mail messages was a Wrong Move.

kai
-- 
~/.signature: No such file or directory

Re: Request: ding subscribers' group parameter

[Lars Magne Ingebrigtsen]

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

> I don't know.  I thought that MCT was for news only, and using it in
> mail messages was a Wrong Move.

I think it's a Good Move.  :-)

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

Re: Request: ding subscribers' group parameter

[Karl Kleinpaste]

prj@po.cwru.edu (Paul Jarc) writes:
> Since Gnus now respects Mail-Followup-To, it'd be better to encourage
> people to add that field to outgoing messages.

Let me see if I've got this straight:

Gnus provides a feature explicitly designed¹ to eliminate duplicate
copies to individuals inhabiting mailing lists.  This feature works
every single time, in every single instance, if only the list
subscribers will deploy it as intended.  Yet we should ignore this
feature, instead in favor of Yet Another Header.

Right.  Just checking.

> I think the best way
> to do this would be to look at the current group's parameters; if
> there is a to-address or to-list parameter and its address is included
> in To+Cc, then we should generate MFT.

"If there is a to-address" in the group parameters...?

If every ding subscriber would use to-address, then the point would
be moot, as the problem would already be solved in the manner suggested.

There is _no reason_ for any ding subscriber /not/ to be using to-address.

I mean, the suggestion was merely to put this valuable Gnus property
to use.  You'd think that Gnus' own development crew would be the very
*first* to put it to use, uniformly and correctly.  It's a no-brainer,
involving the sequence
          M-C-e ((to-address . "ding@gnus.org")) C-c C-c
as you read this message, with minor modifications if you have to
insert the to-address element into a set of other existing parameters.
Instead of everyone just doing that, and the problem disappearing in
the course of a single day, we now have a full-blown discussion over
code changes in order to handle non-existent failure cases in what is
already the default situation when the property is properly in use.

Do I have that right?

If you want to insert MFT headers, fine, but "to-address" provides
exactly the feature you're looking for on *this* list, among *this*
universe of subscribers.  And the price of lunch says MFTs won't do a
bit of good on yahoogroups mailing lists, where Joe Random Users from
all over the planet are still using Outlook Express as packaged with
Win98 which has no awareness of MFT at all.

Solve the problems you can, where you can, as simply as you can.
          M-C-e ((to-address . "ding@gnus.org")) C-c C-c
Do it.  _*/Problem solved./*_

yes, i'm annoyed,
--karl

¹ *Info* -> Gnus -> Group Buffer -> Group Parameters, "to-address":

    "This is primarily useful in mail groups that represent closed
     mailing lists--mailing lists where it's expected that everybody
     that writes to the mailing list is subscribed to it.  Since using
     this parameter ensures that the mail only goes to the mailing list
     itself, it means that members won't receive two copies of your
     followups."

Re: Request: ding subscribers' group parameter

[Kai Großjohann]

On Tue, 31 Jul 2001, Karl Kleinpaste wrote:

> There is _no reason_ for any ding subscriber /not/ to be using
> to-address.

I've always operated on the assumption that people from outside the
list might wish to post.  Does it say somewhere that people from
outside may not post?

kai
-- 
~/.signature: No such file or directory

Re: Request: ding subscribers' group parameter

[Karl Kleinpaste]

Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) writes:
> I've always operated on the assumption that people from outside the
> list might wish to post.  Does it say somewhere that people from
> outside may not post?

No, and people do so from time to time.  The observation is inapposite
to the question at hand: People posting from outside the list are not
following up (badly) to those of us on the list.

Re: Request: ding subscribers' group parameter

[Kai Großjohann]

On Tue, 31 Jul 2001, Karl Kleinpaste wrote:

> Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) writes:
>
>> I've always operated on the assumption that people from outside the
>> list might wish to post.  Does it say somewhere that people from
>> outside may not post?
> 
> No, and people do so from time to time.  The observation is
> inapposite to the question at hand: People posting from outside the
> list are not following up (badly) to those of us on the list.

`Inapposite'.  New word.  Nifty.

I do think the observation is pertinent to the issue at hand,
however.  For if John Smith who is not on the list posts and I have
ding@gnus.org in the to-address/to-list parameter (I always forget
which one), then John will not receive a copy of my answer.

I presume that John would like a copy of my answer, since sending his
question directly to the bit-bucket would be much cheaper and faster
and achieve the same result.

Or am I missing something, here?

kai
-- 
~/.signature: No such file or directory

Re: Request: ding subscribers' group parameter

[Karl Kleinpaste]

Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) writes:
> For if John Smith who is not on the list posts and I have
> ding@gnus.org in the to-address/to-list parameter (I always forget
> which one), then John will not receive a copy of my answer.

Go back through the last, say, 5000 ding postings.  Determine how many
of them were from people not subscribers to the list.  I'll be alarmed
if the number is even in two digits -- yes, sure, it happens, once
every several geological epochs.

Compare that vanishingly small number to e.g. the number of duplicates
I've seen in just the last couple weeks, 7 of them today /alone/.

[1545] [18:48:26] cinnamon:~/Mail/list/ding> grep -l ^Gnus-Warning * | xargs grep ^Date:
38122:Date: 21 Jun 2001 11:46:06 -0700
38428:Date: Fri, 20 Jul 2001 10:27:10 -0500
38447:Date: Sat, 21 Jul 2001 20:21:27 +0200
38460:Date: Sat, 21 Jul 2001 21:07:21 -0500
38468:Date: Sun, 22 Jul 2001 11:28:15 +0200
38469:Date: Sun, 22 Jul 2001 11:04:53 +0200
38625:Date: Fri, 27 Jul 2001 22:20:34 +0200
38626:Date: Sat, 28 Jul 2001 00:35:12 +0200
38744:Date: Tue, 31 Jul 2001 00:39:23 +0200
38745:Date: Tue, 31 Jul 2001 00:37:53 +0200
38765:Date: Tue, 31 Jul 2001 10:56:57 +0200
38770:Date: Tue, 31 Jul 2001 16:44:00 +0200
38783:Date: Tue, 31 Jul 2001 21:57:46 +0200
38784:Date: Tue, 31 Jul 2001 22:11:10 +0200
38791:Date: Wed, 01 Aug 2001 00:44:23 +0200

I use the manual-recommended duplicates-catcher in nnmail-split-rules,
because most of the time, by far, duplicates are just that, and
usually spam -- my duplicates group is set to score down everything
and expire instantly.  For the ding list, I have to move them all back
to nnml:list.ding manually.  Oddly, the ding list is the only mailing
list I get where I consistently see duplicates at all.

For John Smith who posts but is not on the list, `C-c C-t' solves the
problem entirely, as the wildly exceptional, outlying data point in an
otherwise consistent graph enforced by the presence of to-address.

Re: Request: ding subscribers' group parameter

[Nevin Kapur]

Karl Kleinpaste <karl@charcoal.com> writes:

> I use the manual-recommended duplicates-catcher in nnmail-split-rules,
> because most of the time, by far, duplicates are just that, and
> usually spam -- my duplicates group is set to score down everything
> and expire instantly.  For the ding list, I have to move them all back
> to nnml:list.ding manually.  Oddly, the ding list is the only mailing
> list I get where I consistently see duplicates at all.

Can you really call this "manual-recommended"? Yes, it is in the
manual but it is for the express purposes of collecting all duplicate
mail in a separate folder. If you have a split rule for the ding list
and you let email fall through (i.e. not have the duplicate rule)
would not all the ding duplicates land up in the ding folder?

-- 
Nevin

Re: Request: ding subscribers' group parameter

[Steve Youngs]

|--==> "KG" == Kai Großjohann <Kai.Grossjohann@CS.Uni-Dortmund.DE> writes:

  KG> I do think the observation is pertinent to the issue at hand,
  KG> however.  For if John Smith who is not on the list posts and I have
  KG> ding@gnus.org in the to-address/to-list parameter (I always forget
  KG> which one), then John will not receive a copy of my answer.

  KG> I presume that John would like a copy of my answer, since sending his
  KG> question directly to the bit-bucket would be much cheaper and faster
  KG> and achieve the same result.

  KG> Or am I missing something, here?

Yep.  You're forgetting that John, the polite sole that he is, would
include at the start of his post "I'm not subscribed to this list,
could you please Cc me directly, thank you."

And on seeing those words, Kai would 'F' and then 'C-T'.

IMO, anyone who posts to a list that they aren't subscribed to and
doesn't include something similar to the above, doesn't deserve to get
a reply.

-- 
|---<Steve Youngs>---------------<GnuPG KeyID: 9E7E2820>---|
|            XEmacs - It's not just an editor.             |
|                    It's a way of life.                   |
|------------------------------------<youngs@xemacs.org>---|

Re: Request: ding subscribers' group parameter

[Paul Jarc]

Steve Youngs <youngs@xemacs.org> writes:
> Yep.  You're forgetting that John, the polite sole that he is, would
> include at the start of his post "I'm not subscribed to this list,
> could you please Cc me directly, thank you."
> 
> And on seeing those words, Kai would 'F' and then 'C-T'.
> 
> IMO, anyone who posts to a list that they aren't subscribed to and
> doesn't include something similar to the above, doesn't deserve to get
> a reply.

Rather than express this intent in the body of the message, John can
express it in the header with Mail-Followup-To.  Gnus will obey MFT in
preference to to-address (no manual action required on the part of
those sending followups), which is good.  But at present, Gnus will
not copy MFT into followups, which is bad - further followups will not
get Cc'ed.  Would you rather trust everyone in the thread to obey the
reminder, and to insert another reminder in their own messages to Cc
the OP?  MFT is simply the automation of (a generalization of) this
task.


paul

Re: Request: ding subscribers' group parameter

[Karl Kleinpaste]

prj@po.cwru.edu (Paul Jarc) writes:
> Gnus will not copy MFT into followups, which is bad - further
> followups will not get Cc'ed.  Would you rather trust everyone in
> the thread to obey the reminder, and to insert another reminder in
> their own messages to Cc the OP?

Your expressed use of MFT amounts to an attempt to generalize
_ongoing_ external participation in a mailing list.

Non-subscribers may send isolated "please help" or "I've found a bug"
messages to a mailing list, which can require specific response to the
originator; but once such a person determines he wants to engage in
_discussion_ on that list, he can darn well subscribe like the rest of
us mere mortals, participating as subscribers are intended.

At that point, of course, MFT again becomes unnecessary, since
the to-address property controls.

The entire issue of outsiders sending occasional notes to a mailing
list is a vast red herring in the first place.  It happens so rarely
as to be essentially an ignorable case; `C-c C-t' works just fine.

Re: Request: ding subscribers' group parameter

[Paul Jarc]

Karl Kleinpaste <karl@charcoal.com> writes:
> prj@po.cwru.edu (Paul Jarc) writes:
>> Gnus will not copy MFT into followups, which is bad - further
>> followups will not get Cc'ed.  Would you rather trust everyone in
>> the thread to obey the reminder, and to insert another reminder in
>> their own messages to Cc the OP?
> 
> Your expressed use of MFT amounts to an attempt to generalize
> _ongoing_ external participation in a mailing list.

That's one use that it generalizes, yes.

> Non-subscribers may send isolated "please help" or "I've found a bug"
> messages to a mailing list, which can require specific response to the
> originator; but once such a person determines he wants to engage in
> _discussion_ on that list, he can darn well subscribe like the rest of
> us mere mortals, participating as subscribers are intended.

Why?  If we can provide the technology to make it easy for someone to
follow just one thread without subscribing, why is that a bad thing?
Especially when no one is asking you to do any of the work?

> At that point, of course, MFT again becomes unnecessary, since
> the to-address property controls.

to-address says "I'm *guessing* that no one listed in From+To+Cc in
the previous message wants an extra copy, so I'll do them a service by
not sending one"; this requires manual intervention or fails
completely when the guess is wrong.

MCT says "I explicitly declare that I don't (or do) want an extra
copy", and Gnus will respect that preference of the sender, if it is
expressed.

MFT can say everything that MCT says and more, all explicitly, and all
by the person who really knows what their preference is.

MCT and MFT are both already respected by Gnus even if to-address is
set.  This is good: we're taking the real preference over someone
else's guess.

I Gnus were to additionally generating MFT for mailing list messages,
that would mean two things: first, followups from others will
automatically respect my MFT on messages I post - others won't have to
do any extra manual configuration to respect my preference.  This will
work even if they're using Mutt or any other client that supports MFT.
Some of us are subscribed to lists other than this one.

Second, Gnus won't discard similar information about others'
preferences.

Do you find any factual errors in the above?  You may not think that
generating MFT is worth your time, but I'm not asking for your time.
I'm willing to code it myself, as soon as I can figure out how.
Hmm... gnus-msg.el seems to be the place to start...


paul

Re: Request: ding subscribers' group parameter

[Nuutti Kotivuori]

Karl Kleinpaste wrote:
> prj@po.cwru.edu (Paul Jarc) writes:
>> Gnus will not copy MFT into followups, which is bad - further
>> followups will not get Cc'ed.  Would you rather trust everyone in
>> the thread to obey the reminder, and to insert another reminder in
>> their own messages to Cc the OP?

[...]

> The entire issue of outsiders sending occasional notes to a mailing
> list is a vast red herring in the first place.  It happens so rarely
> as to be essentially an ignorable case; `C-c C-t' works just fine.

I agree that to-address is quite sufficient for the ding gnus
mailing-list. I personally still don't quite see how the MFT should
really be done, with all the reply-to, MCT and different newsgroup and
mail semantics. But some people live by MFT it seems.

But in any case, shouldn't Gnus do-the-right-thing(tm) with MFT? 
Atleast it should not stop others from benefiting from MFT, even
though one personally might not like it at all.

So therefore I suggest that we work out how MFT should really work in
Gnus and make the support - which is ofcourse a bit irrelevant to this
discussion under this topic.

-- Naked

Re: Request: ding subscribers' group parameter

[Paul Jarc]

Karl Kleinpaste <karl@charcoal.com> writes:
> Gnus provides a feature explicitly designed¹ to eliminate duplicate
> copies to individuals inhabiting mailing lists.  This feature works
> every single time, in every single instance, if only the list
> subscribers will deploy it as intended.

Right.

> Yet we should ignore this feature, instead in favor of Yet Another
> Header.

No.  We should *also* support Yet Another Header, because the header
field is more expressive than to-address, and that extra
expressiveness can be useful.  Have a look at
<URL:http://cr.yp.to/proto/replyto.html>.

> There is _no reason_ for any ding subscriber /not/ to be using to-address.

Someone may post a question here, but not be subscribed to the list.
They might add "Mail-Followup-To: ding@gnus.org, me@here.net" to
explicitly indicate that they want to receive extra copies of messages
in the thread.  If a list member uses to-address, then MFT will
override that for this message, but further followups won't be sent to
the OP unless MFT is preserved in each additional message.  So, yes,
this is a bit fragile, but it does more than to-address is capable of.

OTOH, I think Mail-Copies-To will discourage extra copies (when you
want to discourage them) without interfering with MFT.  And it's
automatically obeyed by everyone else's Gnus, without requiring extra
configuration on their part.  I can't think of any case where
to-address is preferable to MCT.  to-address puts the decision into
the wrong hands - this preference should be expressed by the sender,
not the recipient.

> Instead of everyone just doing that, and the problem disappearing in
> the course of a single day,

You're never going to get everyone to do anything in a single day.

> we now have a full-blown discussion over code changes in order to
> handle non-existent failure cases in what is already the default
> situation when the property is properly in use.

to-address is *not* set by default.  MFT and MCT *are* respected by
default.

> If you want to insert MFT headers, fine, but "to-address" provides
> exactly the feature you're looking for on *this* list, among *this*
> universe of subscribers.

What about non-subscribers?

> And the price of lunch says MFTs won't do a bit of good on
> yahoogroups mailing lists, where Joe Random Users from all over the
> planet are still using Outlook Express as packaged with Win98 which
> has no awareness of MFT at all.

There's nothing to be done in such cases anyway.  You'll get extra
copies from those people regardless of whether you send them any.


paul

Re: Request: ding subscribers' group parameter

[Kai Großjohann]

On Tue, 31 Jul 2001, Paul Jarc wrote:

> MFT is supposed to be generated after editing the message, just
> before sending, and it should contain all the To+Cc addresses (but
> if the user provided one maually, we leave it alone).

Why not generate MFT together with To and Cc when the user hits `F'
(or `S w' or whatever)?

kai
-- 
~/.signature: No such file or directory

Re: Request: ding subscribers' group parameter

[Paul Jarc]

Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) writes:
> Why not generate MFT together with To and Cc when the user hits `F'
> (or `S w' or whatever)?

Because if the user edits To or Cc, they probably want MFT to match.
(In the rarer cases where this is not so, they can insert MFT before
sending (possibly via posting styles, as Colin is doing), and Gnus
will preserve that.)


paul

Re: Request: ding subscribers' group parameter

[Kai Großjohann]

On Tue, 31 Jul 2001, Karl Kleinpaste wrote:

> It will prevent generation of private duplicates when `f'ollowing up
> to the list.

You could put a Mail-Followups-To and/or Mail-Copies-To header in your
postings...

kai
-- 
~/.signature: No such file or directory

Re: Request: ding subscribers' group parameter

[Steve Youngs]

|--==> "KK" == Karl Kleinpaste <karl@charcoal.com> writes:

  KK> Could folks who subscribe here prettyplease put this parameter in
  KK> the group you use for the list?

  KK> (to-address . "ding@gnus.org")

  KK> It will prevent generation of private duplicates when `f'ollowing up
  KK> to the list.

Here here!

And it will also allow you to do neato stuff with
'gnus-mailing-list-mode'.  Well, actually that requires the 'to-list'
parameter, but their related, sorta kinda.

-- 
|---<Steve Youngs>---------------<GnuPG KeyID: 9E7E2820>---|
|            XEmacs - It's not just an editor.             |
|                    It's a way of life.                   |
|------------------------------------<youngs@xemacs.org>---|

gnus-mailing-list-mode (was: Request: ding subscribers' group parameter)

[Henrik Enberg]

Steve Youngs <youngs@xemacs.org> wrote:
> 
> And it will also allow you to do neato stuff with
> 'gnus-mailing-list-mode'.  Well, actually that requires the 'to-list'
> parameter, but their related, sorta kinda.

This is not mentioned anywhere in the manual.  I was able to figure out what
it does by looking at the source, but I only found out about it from Steve's
message.

Henrik
-- 
I always have fun because I'm out of my mind

Re: gnus-mailing-list-mode (was: Request: ding subscribers' group parameter)

[Kai Großjohann]

On Thu, 02 Aug 2001, Henrik Enberg wrote:

> This is not mentioned anywhere in the manual.  I was able to figure
> out what it does by looking at the source, but I only found out
> about it from Steve's message.

Please file a doc bug report.  Thanks.

kai
-- 
~/.signature: No such file or directory

Re: gnus-mailing-list-mode (was: Request: ding subscribers' group parameter)

[Lars Magne Ingebrigtsen]

Henrik Enberg <henrik@enberg.org> writes:

>> And it will also allow you to do neato stuff with
>> 'gnus-mailing-list-mode'.  Well, actually that requires the 'to-list'
>> parameter, but their related, sorta kinda.
>
> This is not mentioned anywhere in the manual. 

Yup.  Julien, could you write a node in the manual to document what
gnus-ml does?

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

Re: Request: ding subscribers' group parameter

[Paul Jarc]

Steve Youngs <youngs@xemacs.org> writes:
> And [to-address] will also allow you to do neato stuff with
> 'gnus-mailing-list-mode'.  Well, actually that requires the 'to-list'
> parameter, but their related, sorta kinda.

Hmm.  Seems like to-address should also turn on g-m-l-m, shouldn't it?
If to-address is set, then the group is almost certainly a mailing
list, and to-list is almost certainly not also set.


paul

Re: Request: ding subscribers' group parameter

[Jochen Küpper]

>>>>> Paul Jarc wrote on Thu, 02 Aug 2001 11:29:11 -0400:
Paul> Steve Youngs <youngs@xemacs.org> writes:
>> And [to-address] will also allow you to do neato stuff with
>> 'gnus-mailing-list-mode'.  Well, actually that requires the 'to-list'
>> parameter, but their related, sorta kinda.

Paul> Hmm.  Seems like to-address should also turn on g-m-l-m,
Paul> shouldn't it?  If to-address is set, then the group is almost
Paul> certainly a mailing list, and to-list is almost certainly not
Paul> also set.

I have a bunch of groups with to-address set that are not mailing
lists. (Mostly individuals that send from different "accounts" but for
which I wanna use their "official" address anyway.)

Greetings,
Jochen
-- 
Einigkeit und Recht und Freiheit                http://www.Jochen-Kuepper.de
    Liberté, Égalité, Fraternité                GnuPG key: 44BCCD8E
        Sex, drugs and rock-n-roll

Re: Request: ding subscribers' group parameter

[Alexandre Oliva]

On Jul 31, 2001, Karl Kleinpaste <karl@charcoal.com> wrote:

> Could folks who subscribe here prettyplease put this parameter in
> the group you use for the list?

> (to-address . "ding@gnus.org")

> It will prevent generation of private duplicates when `f'ollowing up
> to the list.

I actually prefer to be explicitly Cc:ed even when posting to the
list, because my Score rules will highlight messages addressed to me,
so I don't have to explicitly increase the score of the thread and
hope for the best.

Now, if the other party were to decide for me that I'm a subscriber of
the list so I don't need another copy, I might end up missing the
followup while scanning the hundreds of messages of the dozens of
mailing lists I subscribe.

Fortunately for me, the implied value of Mail-Followups-To is the
complete list of addresses appearing in To: and Cc:, so I don't have
to set Mail-Followups-To.  But if someone decides to change the
implied value of the header, by following your suggestion, they will
be doing me a mis-favor.  Perhaps I should be setting
Mail-Followups-To: in all messages I post, so that my e-mail address
is explicitly listed.  Perhaps not.

IMO, the choice of whether to explicitly receive a copy of a followup
belongs to the poster of the original message, not to the poster of
the followup.  That's why Reply-To: the list is considered harmful,
and that's why people came up with Mail-Followups-To:.

Now, you may argue that the implied value of Mail-Followups-To: is
wrong, but then, you have to convince everybody to change every mail
reader so as to do followups differently.  Or you can retain the good
IMO status quo, and start posting messages with Mail-Followups-To: in
which you collect all To: and Cc: addresses in the original message
except your own e-mail address.  Then, you're clearly stating your
intention of not receiving a copy, without imposing a change of
behavior upon everybody else.

Just my R$0,02.

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                  aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist    *Please* write to mailing lists, not to me

Re: Request: ding subscribers' group parameter

[Steve Youngs]

|--==> "AO" == Alexandre Oliva <oliva@lsd.ic.unicamp.br> writes:

  AO> On Jul 31, 2001, Karl Kleinpaste <karl@charcoal.com> wrote:
  >>Could folks who subscribe here prettyplease put this parameter in
  >>the group you use for the list?

  >>(to-address . "ding@gnus.org")

  >>It will prevent generation of private duplicates when `f'ollowing up
  >>to the list.

  AO> I actually prefer to be explicitly Cc:ed even when posting to the
  AO> list, because my Score rules will highlight messages addressed to me,
  AO> so I don't have to explicitly increase the score of the thread and
  AO> hope for the best.

Why not simply increase the score of your Message-ID?  Works well for me.

,----[ C-h C-i gnus RET m scoring RET m followTAB RET ]
|    If you look closely at your own `Message-ID', you'll notice that the
| first two or three characters are always the same.  Here's two of mine:
| 
|      <x6u3u47icf.fsf@eyesore.no>
|      <x6sp9o7ibw.fsf@eyesore.no>
| 
|    So "my" ident on this machine is `x6'.  This can be exploited--the
| following rule will raise the score on all followups to myself:
| 
|      ("references"
|       ("<x6[0-9a-z]+\\.fsf\\(_-_\\)?@.*eyesore\\.no>"
|        1000 nil r))
`----



-- 
|---<Steve Youngs>---------------<GnuPG KeyID: 9E7E2820>---|
|            XEmacs - It's not just an editor.             |
|                    It's a way of life.                   |
|------------------------------------<youngs@xemacs.org>---|

Re: Request: ding subscribers' group parameter

[Alexandre Oliva]

On Aug  3, 2001, Steve Youngs <youngs@xemacs.org> wrote:

> |--==> "AO" == Alexandre Oliva <oliva@lsd.ic.unicamp.br> writes:
AO> I actually prefer to be explicitly Cc:ed even when posting to the
AO> list, because my Score rules will highlight messages addressed to me,
AO> so I don't have to explicitly increase the score of the thread and
AO> hope for the best.

> Why not simply increase the score of your Message-ID?  Works well for me.

Because then I have to wait for my message to show up in the
newsgroup to find out its Message-ID, remember to increase its score,
and hope the replier's MUA will follow the protocol and add the
Message-ID to References: or In-Reply-To:.  This doesn't work well for
me.

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                  aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist    *Please* write to mailing lists, not to me

Re: Request: ding subscribers' group parameter

[Kai Großjohann]

On 03 Aug 2001, Alexandre Oliva wrote:

> Because then I have to wait for my message to show up in the
> newsgroup to find out its Message-ID, remember to increase its
> score, and hope the replier's MUA will follow the protocol and add
> the Message-ID to References: or In-Reply-To:.  This doesn't work
> well for me.

The idea was to find something which is common for all your message
ids.  For example, the user id goes in there somewhere, and together
with your host name, that should be good enough.

kai
-- 
~/.signature: No such file or directory

Re: Request: ding subscribers' group parameter

[Alexandre Oliva]

On Aug  3, 2001, Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) wrote:

> On 03 Aug 2001, Alexandre Oliva wrote:
>> Because then I have to wait for my message to show up in the
>> newsgroup to find out its Message-ID, remember to increase its
>> score, and hope the replier's MUA will follow the protocol and add
>> the Message-ID to References: or In-Reply-To:.  This doesn't work
>> well for me.

> The idea was to find something which is common for all your message
> ids.  For example, the user id goes in there somewhere, and together
> with your host name, that should be good enough.

I see.  But it still doesn't work if the MUA doesn't set References or
In-Reply-To properly.  I receive a fair amount of e-mail for which `^'
doesn't work to get my reply back, which makes me believe that this
simple set-up wouldn't work.

But still, it's likely to be an improvement for my scoring rules,
since sometimes they won't find my e-mail address if it happens to
occur in a continuation line of To: or Cc:, as in:

Cc: someone@somewhere.org, someoneelse@anywhere.com,
    oliva@lsd.ic.unicamp.br

Thanks for the idea!

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                  aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist    *Please* write to mailing lists, not to me

Re: Request: ding subscribers' group parameter

[Paul Jarc]

Alexandre Oliva <oliva@lsd.ic.unicamp.br> writes:
> On Aug  3, 2001, Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) wrote:
>> The idea was to find something which is common for all your message
>> ids.  For example, the user id goes in there somewhere, and together
>> with your host name, that should be good enough.
> 
> I see.  But it still doesn't work if the MUA doesn't set References or
> In-Reply-To properly.

I think you'll find more software currently handles References and
In-Reply-To well than Mail-Followup-To.  You can use both, though.


paul

Re: Request: ding subscribers' group parameter

[Russ Allbery]

Alexandre Oliva <oliva@lsd.ic.unicamp.br> writes:

> Because then I have to wait for my message to show up in the newsgroup
> to find out its Message-ID, remember to increase its score, and hope the
> replier's MUA will follow the protocol and add the Message-ID to
> References: or In-Reply-To:.  This doesn't work well for me.

No, you don't.  All followups to your messages will be caught by the
following scoring rule:

("references"
 ("<or[0-9a-z]+\.fsf\\(_-_\\)?@feijoada\\.lsd\\.ic\\.unicamp\\.br>"
  200 nil r))

Stick it in all.SCORE and you're all set.  If you use more than one
workstation or account, look at your message IDs to form similar patterns
for those machines and add similar entries.

I've been using this technique for four or five years now at least.

-- 
Russ Allbery (rra@stanford.edu)             <http://www.eyrie.org/~eagle/>

Re: Request: ding subscribers' group parameter

[Alexandre Oliva]

On Aug  3, 2001, Russ Allbery <rra@stanford.edu> wrote:

> ("references"
>  ("<or[0-9a-z]+\.fsf\\(_-_\\)?@feijoada\\.lsd\\.ic\\.unicamp\\.br>"
>   200 nil r))

> Stick it in all.SCORE and you're all set.

> I've been using this technique for four or five years now at least.

Thanks, I've been using something very close to this for a few hours,
and I must admit it's far superior to the one that matches my e-mail
addresses in To: and Cc: "head"ers.  However, I've verified that it
doesn't work in two cases: (i) when the replier's MUA is too dumb to
set References: correctly (fortunately this is rare, but I do care
about e-mail I get from some of the people who choose to use such
broken MUAs) and (ii) it fails to highlight e-mail addressed to me in
my misc e-mail folder, which often helps me tell good e-mail from SPAM
that hides the addressees, which is the most common kind of SPAM in my
statistics.


Anyway, is there any reason why gnus doesn't offer "to" and "cc" as
score strings?  I would find this quite useful, and I don't understand
why it hasn't done before.  Is it just because To: and Cc: don't make
much sense for news?

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                  aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist    *Please* write to mailing lists, not to me

Re: Request: ding subscribers' group parameter

[Florian Weimer]

Alexandre Oliva <oliva@lsd.ic.unicamp.br> writes:

> Anyway, is there any reason why gnus doesn't offer "to" and "cc" as
> score strings?

Gnus was a newsreader initially, and even though you might encounter
To: and Cc: headers in news postings, they won't be stored in the
Overview database of the newsserver, so scoring on them would involve
downloading headers, which is expensive.

Of course, To: and Cc: (or stripped versions of it) could be stored in
the nnml overview...

Re: Request: ding subscribers' group parameter

[Alexandre Oliva]

On Aug  4, 2001, Florian Weimer <fw@deneb.enyo.de> wrote:

> Of course, To: and Cc: (or stripped versions of it) could be stored in
> the nnml overview...

D'oh!  I had forgotten non-"head" nnml scoring was done using the nnml
overview.  Yeah, it makes a lot of sense now.  Anyway, to me, it only
makes sense to score on To: and Cc: in e-mail messages, and it's fast
enough to do that with nnml.  It isn't in the case of news, though,
because every single article is downloaded entirely for scoring
purposes before the summary is presented.  So, it's appropriate to
have the To: and Cc: score rules in nnml:all.SCORE, not in all.SCORE.

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                  aoliva@{cygnus.com, redhat.com}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist    *Please* write to mailing lists, not to me

Re: Request: ding subscribers' group parameter

[Kai Großjohann]

On 04 Aug 2001, Alexandre Oliva wrote:

> (ii) it fails to highlight e-mail addressed to me in my misc e-mail
> folder, which often helps me tell good e-mail from SPAM that hides
> the addressees, which is the most common kind of SPAM in my
> statistics.

This is just a workaround: you could set up a split rule which puts
mail that isn't addressed to a mailing list or to you into a special
folder.

I do this, and it catches about 50% of all spam, and there are almost
no false positives.

kai
-- 
~/.signature: No such file or directory

Re: Request: ding subscribers' group parameter

[Karl Kleinpaste]

Alexandre Oliva <oliva@lsd.ic.unicamp.br> writes:
> Anyway, is there any reason why gnus doesn't offer "to" and "cc" as
> score strings?

(setq gnus-extra-headers '(To Cc Newsgroups Keywords)
      nnmail-extra-headers gnus-extra-headers)

Rebuild overviews (M-x nnml-generate-nov-databases RET) to suck such
headers into your nnml overviews from old mail.

Now score on To and Cc using "extra headers".

I e s p To RET <your name> RET
  ^

Re: Request: ding subscribers' group parameter

[Lars Magne Ingebrigtsen]

Karl Kleinpaste <karl@charcoal.com> writes:

> Alexandre Oliva <oliva@lsd.ic.unicamp.br> writes:
> > Anyway, is there any reason why gnus doesn't offer "to" and "cc" as
>> score strings?
>
> (setq gnus-extra-headers '(To Cc Newsgroups Keywords)
>       nnmail-extra-headers gnus-extra-headers)
>
> Rebuild overviews (M-x nnml-generate-nov-databases RET) to suck such
> headers into your nnml overviews from old mail.
>
> Now score on To and Cc using "extra headers".
>
> I e s p To RET <your name> RET
>   ^

I've now added this to the manual as a new node under Scoring.

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

Re: Request: ding subscribers' group parameter

[Paul Jarc]

Alexandre Oliva <oliva@lsd.ic.unicamp.br> writes:
> Perhaps I should be setting Mail-Followups-To: in all messages I
> post, so that my e-mail address is explicitly listed.  Perhaps not.

(It's Mail-Followup-To, not Mail-FollowupS-To.)  If you do that, it
will override others' to-address settings, but (in cases where MFT is
fully supported, not yet true for Gnus) you'll also get Cc'ed for
followups to followups, and so on.  If that's what you want, then go
for it.


paul

Re: Request: ding subscribers' group parameter

[Per Abrahamsen]

Karl Kleinpaste <karl@charcoal.com> writes:

> Could folks who subscribe here prettyplease put this parameter in
> the group you use for the list?
> 
> (to-address . "ding@gnus.org")
> 
> It will prevent generation of private duplicates when `f'ollowing up
> to the list.

I use it, but it is really only intended for closed lists, i.e. lists
where only members can write.  For open lists, the option is
errorphrone, as it require people manually try to deduce or guess
whether the poster is a subscriber or not.  

That is, however, the only solution on the replyers side.  On the
original senders side, there are two other solutions:

(a) Use MFT, as described earlier in the thread.

(b) Use Mail-Copies-To: never, which have the same effect for mail. 

(c) Suppress duplicates.  I use this, duplicates get to a separate
folder, which I used to check once in a while, but I'm just ignoring
now.  It works.

These solutions have several benefits:

1. They offer a fully automatic solution, thus eliminating human
judgement as a source of errors. 

2. They don't involve trying to get other people to solve your
problems, something which often lead to angry words.

3. (c) only.  You often get answers to your message faster than you
would have done otherwise, since they are send directly.

Personally, I tend to use to-address even in open lists, even with the
extra manual work and risk of errors it involves, and the reduced
service for people using (c), as some people get very angry over
duplicates, and even angrier about advice about how they can avoid
it.  I have grown too old for that kind of flamewars.

Nonetheless, I recommend people who are bothered by duplicates to
consider a combination of (a), (b) and (c) for themselves, rather than
advocating an inferior solution for others.