Gnus v5.2 and GNU Emacs 19.31 Win32

Gnus v5.2 and GNU Emacs 19.31 Win32

[Richard Pieri]

-----BEGIN PGP SIGNED MESSAGE-----

Maybe someone can point me at what I am doing wrong, assuming that I am
doing something wrong and this is not a bug somewhere :).

I have Gnus 5.2 up and running (mostly) nicely on a Windows 95 machine,
using the GNU Emacs 19.31 ntemacs prebuilt binary.  Everything works
fine except for incoming mail (I need to hack a few things for popmail
passwords; once I have it working nicely I'll send Lars a real patch for
nnmail).  Even if I put an incoming mail file on the local system and
point nnmail-spool-file to it, Gnus does not move it to the Mail
directory (which was created correctly).

I am thinking that this is a movemail bug, but I am not certain at this
point.  It could be in how Gnus is calling movemail; if so, I cannot see
what is wrong.

-----BEGIN PGP SIGNATURE-----
Version: 2.6.3
Charset: noconv

iQCVAwUBMbcOG56VRH7BJMxHAQFXFgQAwVUU1rfA4AsbmYzsjLLVU3RZGNnjPxqI
v/Cu88Ay1qUqTn+2F05LaQBk9+8s0ETGmG82ZEQ+W7L+iit5Ke/WTV1vhDZQNDKs
FeHFZrl+RP2QqLzLS0jRr9xoEy+++VILxOnM0Ib7odUIsAGCC1/cmANMsN47FmHU
7jNRilm5VYA=
=dD24
-----END PGP SIGNATURE-----
-- 
Richard Pieri/Information Services \ When in doubt, cop an attitude. -A cat's
<ratinox@unilab.dfci.harvard.edu>   \ guide to life
http://www.dfci.harvard.edu/         \ 

Re: Gnus v5.2 and GNU Emacs 19.31 Win32

[Richard Stanton]

>>>>> "RP" == Richard Pieri <ratinox@unilab.dfci.harvard.edu> writes:
>>>>> "RS" == Richard Stanton <stanton@haas.berkeley.edu> writes:

    RS> I run EMACS under OS/2, using both rmail and nnbabyl to read babyl
    RS> format mail every day. I use movemail (which comes with the OS/2
    RS> EMACS port) to move mail from the spool file to my rmail file, and
    RS> never have any trouble at all.

    RP> Why do you not have the same problems?  Because you are running a
    RP> completely different operating system, is why.

Maybe I could have read the original message a little more closely...

    RP> Second problem: end of line characters on a FAT file system are
    RP> different from Unix (and I suspect OS/2), and this causes the regular
    RP> expressions in Gnus (and who knows what else) to fail miserably.

OS/2 end of line is exactly the same as DOS - CRLF rather than the UNIX
LF. That's why I was surprised you had these problems when I didn't. 

Maybe you could recompile/use the movemail that comes with the OS/2
EMACS port.

Richard Stanton

Re: Gnus v5.2 and GNU Emacs 19.31 Win32

[Lars Magne Ingebrigtsen]

Richard Pieri <ratinox@unilab.dfci.harvard.edu> writes:

> system-type for ntemacs (GNU Emacs for Windows 95/NT) is 'windows-nt'.

Ok, I've now added CR stripping to the movemail action when running
under `windows-nt'.

-- 
  "Yes.  The journey through the human heart 
     would have to wait until some other time."

Re: Gnus v5.2 and GNU Emacs 19.31 Win32

[Richard Pieri]

-----BEGIN PGP SIGNED MESSAGE-----

>>>>> "LMI" == Lars Magne Ingebrigtsen <larsi@ifi.uio.no> writes:

LMI> What are the `system-type's for those systems?

system-type for ntemacs (GNU Emacs for Windows 95/NT) is 'windows-nt'.
I have no idea what it is for OS/2.

-----BEGIN PGP SIGNATURE-----
Version: 2.6.3
Charset: noconv

iQCVAwUBMblywp6VRH7BJMxHAQG15wQAnxuNOpvaoDIwGMC00VkJ7f73MvlIGw9R
3rfG7DiySeR9bFvw9/rSBNtWx7+w8RFszJ7SSY0snRCPbuSYU5uF+76c0nUyIzcN
uxxR8GuSsQeMrufvixev1degI+/F55l+O+zVEyL1I70h3rf9nF38Q2qmnCPB7t8R
0kvyr/8LE+c=
=vjai
-----END PGP SIGNATURE-----
-- 
Richard Pieri/Information Services \ Never sleep alone when you can sleep on
<ratinox@unilab.dfci.harvard.edu>   \ someone's face. -A cat's guide to life
http://www.dfci.harvard.edu/         \ 

Re: Gnus v5.2 and GNU Emacs 19.31 Win32

[Richard Pieri]

-----BEGIN PGP SIGNED MESSAGE-----

>>>>> "RS" == Richard Stanton <stanton@haas.berkeley.edu> writes:

RS> I'm confused.

RS> I run EMACS under OS/2, using both rmail and nnbabyl to read babyl
RS> format mail every day. I use movemail (which comes with the OS/2
RS> EMACS port) to move mail from the spool file to my rmail file, and
RS> never have any trouble at all.

Why do you not have the same problems?  Because you are running a
completely different operating system, is why.

First problem: movemail for Windows NT/95, as distributed in ntemacs,
has had the mail file moving functionality disabled due to the fact that
file locking under Samba does not work at all.  With this disabled the
only thing movemail does do is download mail from a POP server.

Second problem: end of line characters on a FAT file system are
different from Unix (and I suspect OS/2), and this causes the regular
expressions in Gnus (and who knows what else) to fail miserably.

-----BEGIN PGP SIGNATURE-----
Version: 2.6.3
Charset: noconv

iQCVAwUBMblyXJ6VRH7BJMxHAQGA9wQAxWCgGBxnYMpRSCrmJnPv8+cW1sx73JA5
dPTBtKoukOrFfAwoDxDqUOetdID84yWKI88Xy1ifKbniSeE4zSTDMo7xSRQunQ/4
Rm9Tj1LM17EQxn6cv1rH7/t8TZst6GajAuSESJwMysZ8NgT8sJlidHv3eIZNv3Mc
jdq80ndTtXA=
=CoEu
-----END PGP SIGNATURE-----
-- 
Richard Pieri/Information Services \ When in doubt, cop an attitude. -A cat's
<ratinox@unilab.dfci.harvard.edu>   \ guide to life
http://www.dfci.harvard.edu/         \ 

Re: Gnus v5.2 and GNU Emacs 19.31 Win32

[Richard Stanton]

    >> Here's a quick hack hook fix; maybe this can be rolled into the actual
    >> nnmail code?

[...]

    Lars> This seems nice; I could add this to that hook by default when running
    Lars> under systems that need it.  What `system-type's would that be?  `emx'
    Lars> (os/2) and, uhm, win95/nt?  What are the `system-type's for those
    Lars> systems?

I'm confused.

I run EMACS under OS/2, using both rmail and nnbabyl to read babyl
format mail every day. I use movemail (which comes with the OS/2 EMACS
port) to move mail from the spool file to my rmail file, and never
have any trouble at all.

Why don't I have the same problems? One possible thing is that on my
system movemail gets mail from a LOCAL spool file, rather than from
the POP host. The POP retrieval is done by another program (poppoll),
which dumps the mail into the local spool file.

Richard Stanton

Re: Gnus v5.2 and GNU Emacs 19.31 Win32

[Lars Magne Ingebrigtsen]

Richard Pieri <ratinox@unilab.dfci.harvard.edu> writes:

> Here's a quick hack hook fix; maybe this can be rolled into the actual
> nnmail code?
> 
> (add-hook 'nnmail-prepare-incoming-hook 'rat:nnmail-prepare-incoming-hook)
> (defun rat:nnmail-prepare-incoming-hook ()
>   "Strip ^M end of line characters from incoming mail file just before splitting."
>   (goto-char (point-min))
>   (while (re-search-forward "\015$" nil t)
>     (delete-backward-char 1)))

This seems nice; I could add this to that hook by default when running
under systems that need it.  What `system-type's would that be?  `emx'
(os/2) and, uhm, win95/nt?  What are the `system-type's for those
systems?

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

Re: Gnus v5.2 and GNU Emacs 19.31 Win32

[Richard Pieri]

-----BEGIN PGP SIGNED MESSAGE-----

>>>>> "ml" == michael lamoureux <lamour@engin.umich.edu> writes:

richard> Hmmm... not the incoming (babyl) mail file created by
richard> movemail (when it pulls a mail file off a POP3 server).
ml> Works for me about once every 15-30 minutes, 8-9 hours a day
ml> (however often that adds up to be). [...]

Hrm.... <fiddle-fiddle>.... <experiment-experiment>...

FUCK!!!

Score another one for end of line character translations.

Here's a quick hack hook fix; maybe this can be rolled into the actual
nnmail code?

(add-hook 'nnmail-prepare-incoming-hook 'rat:nnmail-prepare-incoming-hook)
(defun rat:nnmail-prepare-incoming-hook ()
  "Strip ^M end of line characters from incoming mail file just before splitting."
  (goto-char (point-min))
  (while (re-search-forward "\015$" nil t)
    (delete-backward-char 1)))

-----BEGIN PGP SIGNATURE-----
Version: 2.6.3
Charset: noconv

iQCVAwUBMbhy456VRH7BJMxHAQHoRgP/XKqiKvEKgdUmpgpjg9hnZJa++NQ1Z0Ss
/q+ykEhx6z+kkQMvA4IOlQLrbC0GpAQql4uMIPiI0e1GM1iF6+kdgg+shQlxFYXg
10QvQ38bMG+xoHrf0QVmGdNRfdj6C64beMewqAOxI/vu6sgSoAL+5dKobPzs9pGv
/y+/ulqMlF0=
=fpeE
-----END PGP SIGNATURE-----
-- 
Richard Pieri/Information Services \ When in doubt, cop an attitude. -A cat's
<ratinox@unilab.dfci.harvard.edu>   \ guide to life
http://www.dfci.harvard.edu/         \ 

Re: Gnus v5.2 and GNU Emacs 19.31 Win32

[michael lamoureux]

 "richard" == Richard Pieri <ratinox@unilab.dfci.harvard.edu> writes:

>>>>>> "LMI" == Lars Magne Ingebrigtsen <larsi@ifi.uio.no> writes:

LMI> nnmail in 5.2 parses babyl as well as Unix mbox and MMDF.

richard> Hmmm... not the incoming (babyl) mail file created by
richard> movemail (when it pulls a mail file off a POP3 server).

Works for me about once every 15-30 minutes, 8-9 hours a day (however
often that adds up to be).  I've never had a problem [Solaris 2.4 POP
server, Solaris 2.4 client, XEmacs 19.13, sgnus .80+ & Gnus 5.2.6]


fyi,
Michael

Re: Gnus v5.2 and GNU Emacs 19.31 Win32

[Richard Pieri]

-----BEGIN PGP SIGNED MESSAGE-----

>>>>> "LMI" == Lars Magne Ingebrigtsen <larsi@ifi.uio.no> writes:

LMI> nnmail in 5.2 parses babyl as well as Unix mbox and MMDF.

Hmmm... not the incoming (babyl) mail file created by movemail (when it
pulls a mail file off a POP3 server).  Gnus made a serious mess of
things when it attempted to split it.  Lots of 0-line garbage articles
composed of babyl headers, and then it finally barfed.  Good thing I had
an unmunged crash-box kicking around :).

I am seriously thinking of ripping^H^H^H^H^H^H^Hborrowing Kyle's vm-pop
for this.

-----BEGIN PGP SIGNATURE-----
Version: 2.6.3
Charset: noconv

iQCVAwUBMbhSNJ6VRH7BJMxHAQHreAP9Et3sjE3LTl48PKlPm9LoyZsT2XM/VOzJ
nYT7iDRL5YD4NNUYJ2wU0laZNpNNUS2vyZ7py9GmQP6t2XkHT1UtMYX6mBPWIpjZ
8CLY/I/MbCV0Obb+AwcPK+QjVkZUp1FjzIUmEf9XePN515NQtLyA6/FbySab7f1s
Lb5lk9bapdE=
=EKzr
-----END PGP SIGNATURE-----
-- 
Richard Pieri/Information Services \ Climb your way to the top; that's why the
<ratinox@unilab.dfci.harvard.edu>   \ drapes are there. -A cat's guide to life
http://www.dfci.harvard.edu/         \ 

Re: Gnus v5.2 and GNU Emacs 19.31 Win32

[Richard Pieri]

-----BEGIN PGP SIGNED MESSAGE-----

>>>>> "LMI" == Lars Magne Ingebrigtsen <larsi@ifi.uio.no> writes:

LMI> nnmail in 5.2 parses babyl as well as Unix mbox and MMDF.

Occasionally it does not like Babyl files for incoming mail.  If/when I
get a backtrace I will send it off.

And I found out where the 'file may be corrupted' problem came from.
Gnus expects the incoming mbox file to have Unix end of line characters,
not MS-DOS end of lines.  Those ^M characters mess up the regular
expression parser something fierce.

-----BEGIN PGP SIGNATURE-----
Version: 2.6.3
Charset: noconv

iQCVAwUBMbhO5Z6VRH7BJMxHAQGTpgP9Fz8AzEsRSrF4wNSCV3yA1TNzxTxzPb1Z
XSK7CW59bSCUPpdeoTsu9xvBD+qXIYdld4UexCOTPdCvXvERVW2tQTRYa7SV/s5S
X92qRX4Mq1aB2qC60q+7bUMiGA/8itdwk+Uetk2rx+f3z3XE0suf2H+mZjAufKLE
ik2d6po/Mco=
=m3sL
-----END PGP SIGNATURE-----
-- 
Richard Pieri/Information Services \ Variety is the spice of life: one day
<ratinox@unilab.dfci.harvard.edu>   \ ignore people, the next day annoy
http://www.dfci.harvard.edu/         \ them. -A cat's guide to life

Re: Gnus v5.2 and GNU Emacs 19.31 Win32

[Richard Pieri]

-----BEGIN PGP SIGNED MESSAGE-----

>>>>> "LMI" == Lars Magne Ingebrigtsen <larsi@ifi.uio.no> writes:

>> Babyl file, which is a big pain in the behind since Gnus is not
>> geared to parse a non-Unix (or MMDF) mail file.

LMI> nnmail in 5.2 parses babyl as well as Unix mbox and MMDF.

It does?  Hrm... then why did it complain about the popped file being
corrupt?  I'm too tired to think about it now; I'll tackle it in the
morning when I'm thinking coherently... or thinking at all.

-----BEGIN PGP SIGNATURE-----
Version: 2.6.3
Charset: noconv

iQCVAwUBMbedw56VRH7BJMxHAQEwagP9H4WQalfJADqo1JjLiXu9SNMk/A+4wceX
t2wuKS8loZtF3l61fWM0wA+BLJw7TIV27cGuz1oG99KeNU+55HVVzU5kXj7OTXD4
AZuRb9WCaMzolXtJ1eCZcm6YBtpfB+7kzfHFHKpWdd0VXo6oWFxX+pCmRjiN2PIP
NiS3YRgbWds=
=JvmB
-----END PGP SIGNATURE-----
-- 
Richard Pieri/Information Services \ When you go out in the world, remember:
<ratinox@unilab.dfci.harvard.edu>   \ being placed on a pedestal is a right,
http://www.dfci.harvard.edu/         \ not a priviledge! -A cat's guide to life

Re: Gnus v5.2 and GNU Emacs 19.31 Win32

[Lars Magne Ingebrigtsen]

Richard Pieri <ratinox@unilab.dfci.harvard.edu> writes:

> However, I have now discovered another problem with the ntemacs
> movemail: it converts the incoming mail file (from the POP3 server) to a
> Babyl file, which is a big pain in the behind since Gnus is not geared
> to parse a non-Unix (or MMDF) mail file. 

nnmail in 5.2 parses babyl as well as Unix mbox and MMDF.

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

Re: Gnus v5.2 and GNU Emacs 19.31 Win32

[Andrew Innes]

On 06 Jun 1996 15:57:25 -0400, Richard Pieri <ratinox@unilab.dfci.harvard.edu> said:
>>>>>>"AI" == Andrew Innes <andrewi@harlequin.co.uk> writes:
>
AI>movemail only reads your inbox (only via POP3 on NT/Win95)
>
>That would do it, then.  Gnus relies on movemail to safely copy the
>inbox to a work file before deleting the inbox.  One workaround is to
>use another program when trying to read from a local spool (the mv
>program in the unix95 archive seems to work; the MS-DOS MOVE.EXE does
>not).  If you are doing this on a "live" filesystem, one that an SMTP
>daemon can write to, do not do this, you could loose mail.

Yup - movemail has the ability to read directly from a mail spool file
as well, with all the proper locking code.  However, this method is
disabled on NT/Win95 precisely because Samba (widely used free software
that lets Windows machines access Unix disks over the network) doesn't
appear to implement the necessary locking calls.

>However, I have now discovered another problem with the ntemacs
>movemail: it converts the incoming mail file (from the POP3 server) to a
>Babyl file, which is a big pain in the behind since Gnus is not geared
>to parse a non-Unix (or MMDF) mail file.  Is this the normal behavior
>for movemail?  Is there any way to disable it (I can't find anything in
>the primary Emacs distribution sources that shows how)?  Or is there a
>replacement for movemail that does not act so -- IMO -- brain-damaged?

Yes, movemail is hard-coded to write the output file in Babyl format,
although it could easily be modified and recompiled to output in a
different format (make sure you call it something other than movemail if
you do that, to avoid confusion).

Curiously, movemail only does this conversion when reading from a POP3
server - when reading a mail spool file directly, it just makes a
verbatim copy.

I'm sure Gnus could be made to borrow some of RMAIL's functions for
parsing Babyl files, if necessary, but modifying movemail itself would
be far simpler.

AndrewI

Re: Gnus v5.2 and GNU Emacs 19.31 Win32

[Richard Pieri]

-----BEGIN PGP SIGNED MESSAGE-----

>>>>> "AI" == Andrew Innes <andrewi@harlequin.co.uk> writes:

AI> movemail only reads your inbox (only via POP3 on NT/Win95)

That would do it, then.  Gnus relies on movemail to safely copy the
inbox to a work file before deleting the inbox.  One workaround is to
use another program when trying to read from a local spool (the mv
program in the unix95 archive seems to work; the MS-DOS MOVE.EXE does
not).  If you are doing this on a "live" filesystem, one that an SMTP
daemon can write to, do not do this, you could loose mail.

However, I have now discovered another problem with the ntemacs
movemail: it converts the incoming mail file (from the POP3 server) to a
Babyl file, which is a big pain in the behind since Gnus is not geared
to parse a non-Unix (or MMDF) mail file.  Is this the normal behavior
for movemail?  Is there any way to disable it (I can't find anything in
the primary Emacs distribution sources that shows how)?  Or is there a
replacement for movemail that does not act so -- IMO -- brain-damaged?

-----BEGIN PGP SIGNATURE-----
Version: 2.6.3
Charset: noconv

iQCVAwUBMbc4KZ6VRH7BJMxHAQGHnQP9ESkHIrexc7zyypuLMQgWZ9uWsEpzIMmu
ILNfwenctSi27jeoZGbLSk2GwTjNe3oDcJBkwDiKI0BAgBERpRcCI7veYM8Zq8h7
+AipsRkvnYlOrBnVW88pTAeW9WpcwIRicKNgp0Tt36y6YhDBL5cytjwkjk3a6vGO
IblXtjxartA=
=YLG4
-----END PGP SIGNATURE-----
-- 
Richard Pieri/Information Services \ Variety is the spice of life: one day
<ratinox@unilab.dfci.harvard.edu>   \ ignore people, the next day annoy
http://www.dfci.harvard.edu/         \ them. -A cat's guide to life