Threading

Threading

[Harry Putnam]

Can anyone tell me what I need to do to get threading to handle what
ever is causing the last three messages in the 'slow as molasses' thread
to not thread under Eze's opener.  I see one is a Cc but should'nt they
all still thread under the subject line?

I see there is no "References:" line in them either for some reason.
But I thought gnus could handle that with proper threading variables
set.  Just haven't figured out how yet.

-- 

Harry Putnam  reader@newsguy.com

Re: Threading

[Stephen Zander]

>>>>> "Harry" == Harry Putnam <reader@newsguy.com> writes:
    Harry> I see there is no "References:" line in them either for
    Harry> some reason.  But I thought gnus could handle that with
    Harry> proper threading variables set.  Just haven't figured out
    Harry> how yet.

Using

	(setq gnus-summary-thread-gathering-function
              'gnus-gather-threads-by-subject)

in .gnus should help that.  It's the default, though, so maybe you should
just check what gnus-summary-thread-gathering-function is set to.

FWIW, I've found that gnus-gather-threads-by-subject doesn't always work.
I swapped to gnus-gather-threads-by-reference, but it's not 100% either

YMMV, mainly due to stupid mailer that seem incapable of issuing 'References'
or 'In-Reply-To' headers.

Several passes through the 'Customizing Threading' section of the manual may
also help (I think I'm about due for another one).

-- 
Stephen
---
"Normality is a statistical illusion." -- me

Re: Threading

[Harry Putnam]

Stephen Zander <gibreel@pobox.com> writes:

> >>>>> "Harry" == Harry Putnam <reader@newsguy.com> writes:
>     Harry> I see there is no "References:" line in them either for
>     Harry> some reason.  But I thought gnus could handle that with
>     Harry> proper threading variables set.  Just haven't figured out
>     Harry> how yet.
> 
> Using
> 
> 	(setq gnus-summary-thread-gathering-function
>               'gnus-gather-threads-by-subject)
> 
> in .gnus should help that.  It's the default, though, so maybe you should
> just check what gnus-summary-thread-gathering-function is set to.
> 
> FWIW, I've found that gnus-gather-threads-by-subject doesn't always work.
> I swapped to gnus-gather-threads-by-reference, but it's not 100% either

I tried your above suggestion by running C-x C-e on it and reopened this
group.  Nothing has changed as far as the thread Eze opened except SSrat
posted and it didn't thread either.

Tried 'threads by reference' but it gives an error; Back trace below:
Signaling: (void-functiongnus-gather-threads-by-reference)((([2((([2"Re: 
Canceling with Gnus 5.6.4?" "Lars Magne Ingebrigtsen <larsi@gnus.org>" 
"01 Apr 1998 16:02:03 +0200""<m3n2e5prms.fsf@sparky.gnus.org>"
"<vafogyocn7h.fsf@ramses.cs.uni-dortmund.de>" .  . . . .] ...))))

Heavy snippage of a list of what looks like it might be all posting in
group (a continuation of above)

  gnus-summary-prepare()
  gnus-summary-read-group-1("nnml:ding-mail" nil t nil nil)
  gnus-summary-read-group("nnml:ding-mail" nil t nil nil) 
  gnus-group-read-group(nil t)
  gnus-group-select-group(nil)
* call-interactively(gnus-group-select-group)



-- 

Harry Putnam  reader@newsguy.com

Re: Threading

[Eze Ogwuma]

Harry Putnam <reader@newsguy.com> writes:

> Stephen Zander <gibreel@pobox.com> writes:
> 
> > >>>>> "Harry" == Harry Putnam <reader@newsguy.com> writes:
> >     Harry> I see there is no "References:" line in them either for
> >     Harry> some reason.  But I thought gnus could handle that with
> >     Harry> proper threading variables set.  Just haven't figured out
> >     Harry> how yet.
> > 
> > Using
> > 
> > 	(setq gnus-summary-thread-gathering-function
> >               'gnus-gather-threads-by-subject)
> > 
> > in .gnus should help that.  It's the default, though, so maybe you should
> > just check what gnus-summary-thread-gathering-function is set to.
> > 
> > FWIW, I've found that gnus-gather-threads-by-subject doesn't always work.
> > I swapped to gnus-gather-threads-by-reference, but it's not 100% either
> 
> I tried your above suggestion by running C-x C-e on it and reopened this
> group.  Nothing has changed as far as the thread Eze opened except SSrat
> posted and it didn't thread either.

Believe it or not it seems that messages from Felix Lee are the
problem. They normally don't enter into the thread correctly (only
once in the last 6400 messages) and unless his message is the opening
message in a thread then the others don't thread in after it.

Felix any ideas?

-- 
Eze Ogwuma

Re: Threading

[Felix Lee]

> Believe it or not it seems that messages from Felix Lee are the
> problem. They normally don't enter into the thread correctly (only
> once in the last 6400 messages) and unless his message is the opening
> message in a thread then the others don't thread in after it.

oh cool.  sorry, I haven't been paying attention to this
thread (and the mailing list archive seems to be not there).

I use mh and mh-e for my email, and my first guess is I have
a replcomps or repl.filter that's doing something unexpected
with the In-reply-to: header.

and in fact, that looks like the problem.
nnheader-header-value only returns the first line of a
multiline header value.  and since I have the message-id on
the second line of the in-reply-to header, gnus never sees
the message-id and so can't thread it.

I'll change my repl.filter :)  someone else can decide the
right way to change gnus.
--

Re: Threading

[Harry Putnam]

Felix Lee <flee@teleport.com> writes:


> and in fact, that looks like the problem.
> nnheader-header-value only returns the first line of a
> multiline header value.  and since I have the message-id on
> the second line of the in-reply-to header, gnus never sees
> the message-id and so can't thread it.
> 
> I'll change my repl.filter :)  someone else can decide the
> right way to change gnus.

I feel like I'm getting off too cheap.  For once it wasn't me doing
something dumb.

One other item that is happening in this thread is that some of the
messages are Cc'ed and so I get two of them.  I wondered if I am also
sending two when I hit 'F' (I'm using the ding mail list not NNTP
server) I've had this 'wide' reply business explained to me a few times
and my understanding is that hitting 'F' on a mail group will send to
the list only and not the author, assuming you've entered the list
address in 'G c' for that group.

Now, I've also seen it said that 'F', in a mail group, replys to
anything in the Cc field too.  Is that right? And so, am I sending
replys to the list and to any thing in the Cc field?


-- 

Harry Putnam  reader@newsguy.com

Re: Threading

[Eze Ogwuma]

Felix Lee <flee@teleport.com> writes:

> > Believe it or not it seems that messages from Felix Lee are the
> > problem. They normally don't enter into the thread correctly (only
> > once in the last 6400 messages) and unless his message is the opening
> > message in a thread then the others don't thread in after it.
> 
> oh cool.  sorry, I haven't been paying attention to this
> thread (and the mailing list archive seems to be not there).

Noticed that yesterday.

[...]

> I'll change my repl.filter :)  someone else can decide the
> right way to change gnus.

Actually a change to Gnus would be good. I see this threading problem
on the Exmh list and also on the Red Hat lists when a RH employee
posts to a thread (Exmh is the mail reader of choice at RH).

-- 
Eze Ogwuma

Re: Threading

[Jason L Tibbitts III]

>>>>> "FL" == Felix Lee <flee@teleport.com> writes:

FL> [...] the mailing list archive seems to be not there).

Which list archive would this be?  The one you get with 'G a', or some
other archive not run at my site?  (I just want to make sure I haven't
broken something.)

 - J<

Re: Threading /

[Felix Lee]

the one at http://www.gnus.org/resources.html, isn't there.

Jason L Tibbitts III <tibbs@hpc.uh.edu>:
> Which list archive would this be?  The one you get with 'G a', or some
> other archive not run at my site?  (I just want to make sure I haven't
> broken something.)

Re: Threading

[Justin Sheehy]

Felix Lee <flee@teleport.com> writes:

> I use mh and mh-e for my email, and my first guess is I have
> a replcomps or repl.filter that's doing something unexpected
> with the In-reply-to: header.

MH is the culprit, but it isn't your In-reply-to: header.  MH, with
its default configuration, adds a space to the end of Subject lines of
messages that it replies to.  After examining your messages, this is
indeed the case.

MH is broken in this regard.

Eze Ogwuma <typhoon@dircon.co.uk> writes:

> Actually a change to Gnus would be good. 

Gnus threads them just fine for me.

Read the docs for gnus-summary-gather-subject-limit.

-- 
Justin Sheehy

In a cloud bones of steel.
  

Re: Threading

[Felix Lee]

Justin Sheehy <justin@linus.mitre.org>:
> MH is the culprit, but it isn't your In-reply-to: header.  MH, with
> its default configuration, adds a space to the end of Subject lines of
> messages that it replies to.  After examining your messages, this is
> indeed the case.

hmm.  yes.  but this is something in the guts of MH, not the
configuration.  the %{subject} format spec has the subject
with a space at the end.  no idea why.

gnus could strip trailing spaces from the subject before
threading.
--

Re: Threading

[Justin Sheehy]

Felix Lee <flee@teleport.com> writes:

> but this is something in the guts of MH, not the configuration.

The default is indeed broken in mh, but a user that wants to can force
correct behavior.  Perhaps using %(trim) in the replcomps file?  I
don't remember, it's been ages since I used mh.

> gnus could strip trailing spaces from the subject before
> threading.

It already can.

(setq gnus-summary-gather-subject-limit 'fuzzy) and similar things can
be pretty handy.

-- 
Justin Sheehy

In a cloud bones of steel.
  

Re: Threading

[Felix Lee]

Justin Sheehy <justin@linus.mitre.org>:
> The default is indeed broken in mh, but a user that wants to can force
> correct behavior.  Perhaps using %(trim) in the replcomps file?  I
> don't remember, it's been ages since I used mh.

yah.  replace %{subject} in replcomps with %(putstr(trim{subject}))

still, it's not the default, and it's far from an obviously
necessary change, and it'll mess up subjects that
intentionally have trailing spaces for whatever reason.
--

Re: Threading

[Dave Love]

>>>>> "Harry" == Harry Putnam <reader@newsguy.com> writes:

 Harry> Can anyone tell me what I need to do to get threading to
 Harry> handle what ever is causing the last three messages in the
 Harry> 'slow as molasses' thread to not thread under Eze's opener.

In a Gnus with the fix for grokking In-reply-to:, adding
`gnus-simplify-whitespace' to `gnus-simplify-subject-functions' is
sufficient to thread MH stuff.  At least, I think that's what in my
setup does the trick.

[Some news client does the same thing with subject lines, but I don't
know which.]

Re: Threading

[Mike McEwan]

Harry Putnam <reader@newsguy.com> writes:

> Can anyone tell me what I need to do to get threading to handle what
> ever is causing the last three messages in the 'slow as molasses' thread
> to not thread under Eze's opener.  I see one is a Cc but should'nt they
> all still thread under the subject line?

  Here's what I have in my `gnus.el' :

(add-hook 'gnus-summary-generate-hook
	  '(lambda ()
	     (if (or (not (gnus-news-group-p gnus-newsgroup-name))
			  (equal gnus-newsgroup-name
                                 "nntp+sunsite.auc.dk:sunsite.linux.kernel"))
		 (setq gnus-simplify-subject-functions
		       '(gnus-simplify-subject-re
			 gnus-simplify-whitespace))
	       (setq gnus-simplify-subject-functions nil))))

  This effectively ignores the `Re' that begins subject lines and the
whitespace that ends them, when the group in question is not a
bona-fide newsgroup. The reason for the `sunsite.auc.dk' reference is
that, although I take it in over nntp, it is really a gateway-ed
mailing list, and as such, articles do not always have `references'
lines that gnus can easily thread by.

  `References' is the header line that, as I understand it, form real
newsgroup threads - anything else is really guess-work on the part of
the newsreader. At least gnus gives us a choice of how we wish to
relate articles in the absence of such criteria.

HTH
-- 
Mike.