pgnus on a Windows box

pgnus on a Windows box

[Jack Vinson]

[-- Attachment #1: Type: text/plain, Size: 552 bytes --]

Okay, I am sure there is an easy answer for this, but what variable to I
have to set to fix this?

When I open my old archives (from Gnus 5.6.55) pgnus complains.  If I look
at the file (nnfolder+archive) each line has a \r appended to the end.  I
assume this has something to do with the fact that I am on Windows 95
machine.

If I do a query replace to remove all the \r's, I am able to enter the
group.  However, when I leave the group the file apparently closes and
reentry gives the same problem.  The backtrace is included below for
completeness.


[-- Attachment #2: Type: text/plain, Size: 3682 bytes --]

Signaling: (args-out-of-range 1 0)
  insert-buffer-substring(#<buffer misc-mail> 1 0)
  nnfolder-retrieve-headers((641 642 645 648 652 654 655 656 657 658 659 660 661 662 664 665 666 667 669 670 671 672 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 694 696 697 698 699 700 701 702 703 ... . (704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 761)) "misc-mail" "archive" some)
  gnus-retrieve-headers((641 642 645 648 652 654 655 656 657 658 659 660 661 662 664 665 666 667 669 670 671 672 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 694 696 697 698 699 700 701 702 703 ... . (704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 761)) "nnfolder+archive:misc-mail" some)
  gnus-cache-retrieve-headers((641 642 645 648 652 654 655 656 657 658 659 660 661 662 664 665 666 667 669 670 671 672 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 694 696 697 698 699 700 701 702 703 ... . (704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 761)) "nnfolder+archive:misc-mail" some)
  gnus-retrieve-headers((641 642 645 648 652 654 655 656 657 658 659 660 661 662 664 665 666 667 669 670 671 672 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 694 696 697 698 699 700 701 702 703 ... . (704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 761)) "nnfolder+archive:misc-mail" some)
  gnus-select-newsgroup("nnfolder+archive:misc-mail" nil nil)
  gnus-summary-read-group-1("nnfolder+archive:misc-mail" nil nil nil nil nil)
  gnus-summary-read-group("nnfolder+archive:misc-mail" nil nil nil nil nil nil)
  gnus-group-read-group(nil nil nil)
  gnus-topic-read-group(nil)
  call-interactively(gnus-topic-read-group)
  completing-read("Describe function: " [mail-extr-x400-encoded-address-surname-pattern mail-extr-hz-embedded-gb-encoded-chinese-pattern bface bmrk-filename gnus-signature-toggle last-time cl-seq1 cl-seq2 start-char gnus-add-to-list bbdb-refile-notes-default-merge-function bbdb/gnus-summary-prefer-bbdb-data View-quit gnus-article-summary-command-nosave gnus-topic-get-new-news-this-topic names-match-flag sort-columns-subprocess bookmark-bmenu-surreptitiously-rebuild-list bmrk bookmark-set-filename will-go nnmail-search-unix-mail-delim-backward end-line pop3-read-passwd widget-insert gnus-cite-max-prefix gnus-group-topic-name buffer-elisp cl-keys-temp epop3-parse-po:user@host gcc kinsoku-longer widget-link-prefix cl-copy gnus-cite-hide-absolute gnus-decode-header-methods ange-ftp-parse-netrc-token ispell-lookup-last-bow yanked view-mode-hook temp-file-name ispell-next-message gnus-article-signature widget-editable-list-insert-before supported-extensions smtpmail-recipient-addr!
 
ess-list convert-widget widget-doc
umentation-string-action gnus-summary-set-saved-mark gnus-orphan-score] fboundp t nil nil "nil")
  (setq val (completing-read (if fn ... "Describe function: ") obarray (quote fboundp) t nil nil (symbol-name fn)))
  (let ((fn ...) (enable-recursive-minibuffers t) val) (setq val (completing-read ... obarray ... t nil nil ...)) (list (if ... fn ...)))
  call-interactively(describe-function)

[-- Attachment #3: Type: text/plain, Size: 152 bytes --]


-- 
Jack Vinson <jvinson@chevax.ecs.umass.edu>    http://www.cis.upenn.edu/~vinson/
Zippy: TATTOOED MIDGETS are using ALFREDO in their SALAMI FACTORY!

Re: pgnus on a Windows box

[Shenghuo ZHU]

>>>>> "Jack" == Jack Vinson <jvinson@chevax.ecs.umass.edu> writes:

Jack> Okay, I am sure there is an easy answer for this, but what variable to I
Jack> have to set to fix this?

Jack> When I open my old archives (from Gnus 5.6.55) pgnus complains.  If I look
Jack> at the file (nnfolder+archive) each line has a \r appended to the end.  I
Jack> assume this has something to do with the fact that I am on Windows 95
Jack> machine.

Jack> If I do a query replace to remove all the \r's, I am able to enter the
Jack> group.  However, when I leave the group the file apparently closes and
Jack> reentry gives the same problem.  The backtrace is included below for
Jack> completeness.

Try (setq nnmail-file-coding-system 'raw-text).

But it may lead other errors.

-- 
Shenghuo

Re: pgnus on a Windows box

[Jack Vinson]

>>>>> "ZSH" == Shenghuo ZHU <zsh@cs.rochester.edu> writes:

>>>>> "Jack" == Jack Vinson <jvinson@chevax.ecs.umass.edu> writes:
[snip]
Jack> When I open my old archives (from Gnus 5.6.55) pgnus complains.  If I
Jack> look at the file (nnfolder+archive) each line has a Ctrl-M appended
Jack> to the end. 
[snip]

ZSH> Try (setq nnmail-file-coding-system 'raw-text).

ZSH> But it may lead other errors.

Yes, it does.  I am not including the backtrace this time, but a different
error occurs when I use this coding system.  Drat.  

I've also noticed that old mail messages (nnml) that were saved before the
transition to Emacs 20.3 are a little strange.  Pgnus is not able to parse
the header information out of the file.  Actually, it only seems to be the
code associated with Message.  The summary buffer looks right, but replying
or following up to a message gives me a message buffer with an empty
Subject and no To line.  It has no problems with messages that have been
stored since the arrival of 20.3.

-- 
Jack Vinson <jvinson@chevax.ecs.umass.edu>    http://www.cis.upenn.edu/~vinson/
Zippy: BI-BI-BI-BI-BI-BI-BI-BI-BI-BI-BI-BI-BI-BI-BI-BI-BI-BI-BI-BI-BI-BI-BI-BI-BI-BI-

Re: pgnus on a Windows box

[Shenghuo ZHU]

>>>>> "Jack" == Jack Vinson <jvinson@chevax.ecs.umass.edu> writes:

>>>>> "ZSH" == Shenghuo ZHU <zsh@cs.rochester.edu> writes:
>>>>> "Jack" == Jack Vinson <jvinson@chevax.ecs.umass.edu> writes:
Jack> [snip]
Jack> When I open my old archives (from Gnus 5.6.55) pgnus complains.  If I
Jack> look at the file (nnfolder+archive) each line has a Ctrl-M appended
Jack> to the end. 
Jack> [snip]

ZSH> Try (setq nnmail-file-coding-system 'raw-text).

ZSH> But it may lead other errors.

Jack> Yes, it does.  I am not including the backtrace this time, but a different
Jack> error occurs when I use this coding system.  Drat.  

Jack> I've also noticed that old mail messages (nnml) that were saved before the
Jack> transition to Emacs 20.3 are a little strange.  Pgnus is not able to parse
Jack> the header information out of the file.  Actually, it only seems to be the
Jack> code associated with Message.  The summary buffer looks right, but replying
Jack> or following up to a message gives me a message buffer with an empty
Jack> Subject and no To line.  It has no problems with messages that have been
Jack> stored since the arrival of 20.3.

Try this patch and (setq nnmail-file-coding-system 'raw-text). Please
report whether it works or not.

-- 
Shenghuo

:- cut ---------------
--- ChangeLog	1998/11/19 20:22:24	1.1
+++ ChangeLog	1998/11/19 20:26:57
@@ -1,3 +1,9 @@
+Thu Nov 19 15:26:02 1998  Shenghuo ZHU  <zsh@cs.rochester.edu>
+
+	* nnmail.el (nnmail-nov-file-coding-system): New variable.
+	* nnml.el (nnml-open-nov): Use nnmail-nov-file-coding-system.
+	(nnml-save-nov): Use nnmail-nov-file-coding-system.
+
 Thu Nov 19 04:48:42 1998  Lars Magne Ingebrigtsen  <larsi@menja.ifi.uio.no>
 
 	* gnus.el: Pterodactyl Gnus v0.51 is released.

--- nnmail.el	1998/11/19 20:19:49	1.1
+++ nnmail.el	1998/11/19 20:20:20
@@ -496,6 +496,9 @@
 (defvar nnmail-file-coding-system 'binary
   "Coding system used in nnmail.")
 
+(defvar nnmail-nov-file-coding-system 'binary
+  "Coding system used in nov file.")
+
 (defun nnmail-find-file (file)
   "Insert FILE in server buffer safely."
   (set-buffer nntp-server-buffer)

--- nnml.el	1998/11/19 20:16:43	1.1
+++ nnml.el	1998/11/19 20:21:06
@@ -686,7 +686,8 @@
 		       nnml-nov-file-name))
 	  (erase-buffer)
 	  (when (file-exists-p nnml-nov-buffer-file-name)
-	    (nnheader-insert-file-contents nnml-nov-buffer-file-name)))
+	    (let ((nnheader-file-coding-system nnmail-nov-file-coding-system))
+	      (nnheader-insert-file-contents nnml-nov-buffer-file-name))))
 	(push (cons group buffer) nnml-nov-buffer-alist)
 	buffer)))
 
@@ -696,8 +697,9 @@
       (when (buffer-name (cdar nnml-nov-buffer-alist))
 	(set-buffer (cdar nnml-nov-buffer-alist))
 	(when (buffer-modified-p)
-	  (nnmail-write-region 1 (point-max) nnml-nov-buffer-file-name
-			       nil 'nomesg))
+	  (let ((nnmail-file-coding-system nnmail-nov-file-coding-system))
+	    (nnmail-write-region 1 (point-max) nnml-nov-buffer-file-name
+				 nil 'nomesg)))
 	(set-buffer-modified-p nil)
 	(kill-buffer (current-buffer)))
       (setq nnml-nov-buffer-alist (cdr nnml-nov-buffer-alist)))))

Re: pgnus on a Windows box

[Jack Vinson]

>>>>> "ZSH" == Shenghuo ZHU <zsh@cs.rochester.edu> writes:

ZSH> Try this patch and (setq nnmail-file-coding-system 'raw-text). Please
ZSH> report whether it works or not.

The patch Shenghou created didn't help.  The problem is obviously the
coding system.  I tried 'raw-text-dos (as opposed to 'raw-text) with the
original files and that seemed to do the trick.  To read my nnfolders I had
to only set nnheader-file-coding-system.

There are other problems:
	* drafts get saved with the \r and Gnus forgets to ignore them when
	  loading the file.  I need to set nnmail-file-coding-system to
	  'raw-text-dos to get this right.
	* MIME binary parts (jpeg & gif) are getting saved with trailing
	  \r's and I can't view the image.  It looks like the
	  buffer-file-coding-system is getting set to 'undecided-dos which
	  happens to be the default-buffer-file-coding-system.
	* Message doesn't quite understand mail saved before I upgraded
	  to 20.3.  When I reply to older mail, message is unable to find
	  the author and subject.  The summary buffer displays this
	  information with no problem.

There MUST be an easier way to deal with all of this.  I'll ask the ntemacs
gang for help.

-- 
Jack Vinson <jvinson@chevax.ecs.umass.edu>    http://www.cis.upenn.edu/~vinson/
Zippy: -- In 1962, you could buy a pair of SHARKSKIN SLACKS, with a
 ``Continental Belt,'' for $10.99!!

Re: pgnus on a Windows box

[Shenghuo ZHU]

>>>>> "Jack" == Jack Vinson <jvinson@chevax.ecs.umass.edu> writes:

>>>>> "ZSH" == Shenghuo ZHU <zsh@cs.rochester.edu> writes:
ZSH> Try this patch and (setq nnmail-file-coding-system 'raw-text). Please
ZSH> report whether it works or not.

Jack> The patch Shenghou created didn't help.  The problem is
Jack> obviously the coding system.  I tried 'raw-text-dos (as opposed
Jack> to 'raw-text) with the original files and that seemed to do the

It is weird. raw-text should be equivalent to raw-text-dos in NTEmacs.
Maybe it is a bug in NTEmacs.

DOC> t -- raw-text
DOC>   Raw text, which means text contains random 8-bit codes.
DOC> Type: 5 (text with random binary characters)
DOC> EOL type: Automatic selection from:
DOC> 	[raw-text-unix raw-text-dos raw-text-mac]
DOC> This coding system can encode charsets:

Jack> trick.  To read my nnfolders I had to only set
Jack> nnheader-file-coding-system.

It seems that gnus needs more coding-system variables. Each backend
needs a file-coding-system and path-coding-system, since backends may
locate on different file systems and share with other os.


Jack> There are other problems:
Jack> 	* drafts get saved with the  and Gnus forgets to ignore them when
Jack> 	  loading the file.  I need to set nnmail-file-coding-system to
Jack> 	  'raw-text-dos to get this right.

Could you test my patch in "PATCH: more charset"? And remember to set
"charset" to "nil" in nndraft:drafts ( `G c' on nndraft:drafts) or
delete (`C-k') it and gnus will recreate it next time.

Jack> 	* MIME binary parts (jpeg & gif) are getting saved with trailing
Jack> 	  's and I can't view the image.  It looks like the
Jack> 	  buffer-file-coding-system is getting set to 'undecided-dos which
Jack> 	  happens to be the default-buffer-file-coding-system.

Could you test my patch in "Re: MIME decoding broken on Win NT"?

Jack> 	* Message doesn't quite understand mail saved before I upgraded
Jack> 	  to 20.3.  When I reply to older mail, message is unable to find
Jack> 	  the author and subject.  The summary buffer displays this
Jack> 	  information with no problem.

Weird. If possible, could you send a (encoded) sample?

-- 
Shenghuo

Re: pgnus on a Windows box

[Jack Vinson]

[-- Attachment #1: Type: text/plain, Size: 1189 bytes --]

>>>>> "ZSH" == Shenghuo ZHU <zsh@cs.rochester.edu> writes:

Jack> * Message doesn't quite understand mail saved before I upgraded
Jack> to 20.3.  When I reply to older mail, message is unable to find
Jack> the author and subject.  The summary buffer displays this
Jack> information with no problem.

ZSH> Weird. If possible, could you send a (encoded) sample?

Sure, I can try.  The problem is completely solved if I do
(setq nnmail-file-coding-system 'raw-text-dos)

But the example is pretty straightforward.  With this value set to 'binary,
my summary buffer looks normal. As shown in the enclosed ex_summary.txt.
(All of these messages happen to have come before I moved to emacs-20.3.1.)
When I hit reply on a message, the message buffer looks like
ex_message1.txt.  If I change nnmail-file-coding-system to 'raw-text-dos
and reply to the same message, the message buffer looks correct as shown in
ex_message2.txt.  For this to work I must select a different message and
then reselect the first message for the nnmail-file-coding-system to take
effect when it reads that file.  (I am working in the nnml back end.)

I have also set
(setq nnheader-file-coding-system 'no-conversion)



[-- Attachment #2: Type: text/plain, Size: 699 bytes --]

!  [ 232: *Onsale Auto Reply  ] Congratulations! Your bid #5137130 was successful.
O  [  80: *Onsale Auto Reply  ] Your recent bid has been reinstated as a currently winning bid
!  [ 216: *Onsale Auto Reply  ] Congratulations! Your bid #5154369 was successful.
!  [ 182: *Onsale Auto Reply  ] Your Bid #5438668 received
!  [ 206: *Onsale Auto Reply  ] Congratulations! Your bid #5438668 was successful.
E  [  43: *Onsale Maillist    ] New at ONSALE - Expanded Quick Buy
E  [  61: *Onsale Maillist    ] Weekend Monitor Madness with bids starting at $1
E  [  59: *Onsale Maillist    ] Sony Hi Res Advanced Digital Camera starts @ $1
E  [  66: *Onsale Maillist    ] 11 ProForm Treadmills starting @ $1

[-- Attachment #3: Type: text/plain, Size: 116 bytes --]

Subject: Re: none
Gcc: nnfolder+archive:misc-mail
X-Zippy: Look DEEP into the OPENINGS!!
--text follows this line--

[-- Attachment #4: Type: text/plain, Size: 240 bytes --]

To: ONSALE <Onsale_Maillist@new-jersey.onsale.com>
Subject: Re: Weekend Monitor Madness with bids starting at $1
References: <199810162354.TAA22916@unagi.cis.upenn.edu>
Gcc: nnfolder+archive:misc-mail
X-Zippy: ..
--text follows this line--

[-- Attachment #5: Type: text/plain, Size: 111 bytes --]


-- 
Jack Vinson <jvinson@chevax.ecs.umass.edu>    http://www.cis.upenn.edu/~vinson/
Zippy: My EARS are GONE!!

Re: pgnus on a Windows box

[Shenghuo ZHU]

>>>>> "Jack" == Jack Vinson <jvinson@chevax.ecs.umass.edu> writes:

Jack> Sure, I can try.  The problem is completely solved if I do
Jack> (setq nnmail-file-coding-system 'raw-text-dos)

Jack> But the example is pretty straightforward.  With this value set
Jack> to 'binary, my summary buffer looks normal. As shown in the
Jack> enclosed ex_summary.txt.  (All of these messages happen to have
Jack> come before I moved to emacs-20.3.1.)  When I hit reply on a
Jack> message, the message buffer looks like ex_message1.txt.  If I
Jack> change nnmail-file-coding-system to 'raw-text-dos and reply to
Jack> the same message, the message buffer looks correct as shown in
Jack> ex_message2.txt.  For this to work I must select a different
Jack> message and then reselect the first message for the
Jack> nnmail-file-coding-system to take effect when it reads that
Jack> file.  (I am working in the nnml back end.)

Test this patch. Some new voos are added.

-- 
Shenghuo

:- cut ----------------------
--- ChangeLog	1998/11/29 09:55:03	1.2
+++ ChangeLog	1998/11/29 10:04:19
@@ -1,3 +1,11 @@
+Sun Nov 29 04:55:40 1998  Shenghuo ZHU  <zsh@cs.rochester.edu>
+
+	* nnmail.el (nnmail-file-coding-system-1): New variable.
+	* nnfolder.el (nnfolder-file-coding-system): Ditto.
+	(nnfolder-read-folder): Use nnfolder-file-coding-system.
+	* nnml.el (nnml-file-coding-system): New variable.
+	(nnml-request-article): Use nnml-file-coding-system.
+
 Sun Nov 29 03:50:20 1998  Shenghuo ZHU  <zsh@cs.rochester.edu>
 
 	* mm-decode.el (mm-inlinable-p): Save the result of test.

--- nnmail.el	1998/11/29 09:51:30	1.1
+++ nnmail.el	1998/11/29 09:54:36
@@ -496,6 +496,11 @@
 (defvar nnmail-file-coding-system 'binary
   "Coding system used in nnmail.")
 
+(defvar nnmail-file-coding-system-1 
+  (if (string-match "nt" system-configuration)
+      'raw-text-dos 'binary)
+  "Another coding system used in nnmail.")
+
 (defun nnmail-find-file (file)
   "Insert FILE in server buffer safely."
   (set-buffer nntp-server-buffer)

--- nnfolder.el	1998/11/29 09:46:50	1.1
+++ nnfolder.el	1998/11/29 09:57:05
@@ -90,6 +90,7 @@
 (defvoo nnfolder-buffer-alist nil)
 (defvoo nnfolder-scantime-alist nil)
 (defvoo nnfolder-active-timestamp nil)
+(defvoo nnfolder-file-coding-system nnmail-file-coding-system-1)
 
 \f
 
@@ -682,7 +683,10 @@
 
 (defun nnfolder-read-folder (group)
   (let* ((file (nnfolder-group-pathname group))
-	 (buffer (set-buffer (nnheader-find-file-noselect file))))
+	 (buffer (set-buffer 
+		  (let ((nnmail-file-coding-system 
+			 nnfolder-file-coding-system))
+		    (nnheader-find-file-noselect file)))))
     (if (equal (cadr (assoc group nnfolder-scantime-alist))
 	       (nth 5 (file-attributes file)))
 	;; This looks up-to-date, so we don't do any scanning.
--- nnml.el	1998/11/29 09:53:00	1.1
+++ nnml.el	1998/11/29 09:59:55
@@ -86,6 +86,8 @@
 
 (defvar nnml-nov-buffer-file-name nil)
 
+(defvoo nnml-file-coding-system nnmail-file-coding-system-1)
+
 \f
 
 ;;; Interface functions.
@@ -183,7 +185,9 @@
       (nnheader-report 'nnml "No such file: %s" path))
      ((file-directory-p path)
       (nnheader-report 'nnml "File is a directory: %s" path))
-     ((not (save-excursion (nnmail-find-file path)))
+     ((not (save-excursion (let ((nnmail-file-coding-system 
+				  nnml-file-coding-system)) 
+			     (nnmail-find-file path))))
       (nnheader-report 'nnml "Couldn't read file: %s" path))
      (t
       (nnheader-report 'nnml "Article %s retrieved" id)

Re: pgnus on a Windows box

[Jack Vinson]

>>>>> "ZSH" == Shenghuo ZHU <zsh@cs.rochester.edu> writes:

ZSH> Test this patch. Some new voos are added.

This solves the problem of pgnus not quite understanding files that were
saved before I moved to a version of emacs that explicitly deals with
file-coding.

Thanks!

-- 
Jack Vinson <jvinson@chevax.ecs.umass.edu>    http://www.cis.upenn.edu/~vinson/
Zippy: Jesus is my POSTMASTER GENERAL..