[Q] malfunction in nnmail-split-fancy ()

[Q] malfunction in nnmail-split-fancy ()

[Didier Verna]

        I have problems with nnmail-split-fancy. The symptom is that some
articles are split to wrong groups when I retreive them (`g'), but correctly
respooled with `B r'. After tracking down a bit, I notice that when the
articles are retreived and put in the *nntpd* buffer, the buffer doesn't
contain all headers (in my particular case, x-from-line) so my split method
fails. However, once they are displayed, they've got it and I can respool them
normally. So I have three questions:

* how is it that some headers are missing in *nntpd* ?
* are there any variables that control this behavior ?
* if yes, is it really a Good Thing that they influence internal processing ?

-- 
    /     /   _   _       Didier Verna        http://www.inf.enst.fr/~verna/
 - / / - / / /_/ /      E.N.S.T. INF C201.1      mailto:verna@inf.enst.fr
/_/ / /_/ / /__ /        46 rue Barrault        Tel.   (33) 01 45 81 73 46
                      75634 Paris  cedex 13     Fax.   (33) 01 45 81 31 19

Re: [Q] malfunction in nnmail-split-fancy ()

[Kai.Grossjohann]

Well, I think X-From-Line is a special case, because that's the
envelope from header (shows up as the infamous "From " line in the
incoming files).  So maybe you should split on both "From " and
"X-From-Line:"? 
kai
-- 
I like _\bb_\bo_\bt_\bh kinds of music.

Re: [Q] malfunction in nnmail-split-fancy ()

[Didier Verna]

Kai Grossjohann writes:

Kai> Well, I think X-From-Line is a special case, because that's the envelope
Kai> from header (shows up as the infamous "From " line in the incoming
Kai> files). So maybe you should split on both "From " and "X-From-Line:"?

        Actually, I use x-from-line only as a last ressort to split mailing
lists which I couldn't detect otherwise. But the real question is that since
you have the possibility to view those headers in the article buffer (and thus 
respooling based on them), I think they should be seen also when the initial
splitting occurs.


-- 
    /     /   _   _       Didier Verna        http://www.inf.enst.fr/~verna/
 - / / - / / /_/ /      E.N.S.T. INF C201.1      mailto:verna@inf.enst.fr
/_/ / /_/ / /__ /        46 rue Barrault        Tel.   (33) 01 45 81 73 46
                      75634 Paris  cedex 13     Fax.   (33) 01 45 81 31 19

Re: [Q] malfunction in nnmail-split-fancy ()

[Kai.Grossjohann]

Didier Verna <verna@inf.enst.fr> writes:

  > Actually, I use x-from-line only as a last ressort to split
  > mailing lists which I couldn't detect otherwise. But the real
  > question is that since you have the possibility to view those
  > headers in the article buffer (and thus respooling based on them),
  > I think they should be seen also when the initial splitting
  > occurs.

Yes, but the renaming is only a last resort :-)

Have you tried to split on "From "?  How did it turn out?

kai
-- 
I like _\bb_\bo_\bt_\bh kinds of music.

Re: [Q] malfunction in nnmail-split-fancy ()

[Hrvoje Niksic]

Kai.Grossjohann@CS.Uni-Dortmund.DE writes:

>   > Actually, I use x-from-line only as a last ressort to split
[...]
> Have you tried to split on "From "?

Kai, x-from-line is the Gnus way of saying "From ".  (And Real Cool
people spell "From " as From_ anyway.)

Re: [Q] malfunction in nnmail-split-fancy ()

[Kai.Grossjohann]

Hrvoje Niksic <hniksic@srce.hr> writes:

  > Kai, x-from-line is the Gnus way of saying "From ".  (And Real Cool
  > people spell "From " as From_ anyway.)

Well, I had already thought that From_ line doesn't fit the usual
colon header syntax and that it therefore had to be changed.  Question
is, when does that happen?  Before nnmail-split-methods is run or
after?

kai
-- 
I like _\bb_\bo_\bt_\bh kinds of music.

Re: [Q] malfunction in nnmail-split-fancy ()

[Hrvoje Niksic]

Kai.Grossjohann@CS.Uni-Dortmund.DE writes:

> Well, I had already thought that From_ line doesn't fit the usual
> colon header syntax and that it therefore had to be changed.
> Question is, when does that happen?  Before nnmail-split-methods is
> run or after?

It used to be that it was converted *after* nnmail-split-methods are
run, but I haven't checked recently.  I clearly recall being
frustrated about not being able to match From_ with nnmail-split-fancy 
because it insisted on adding the `:' stuff to the headers, while not
providing for the From_ anomaly.