pgnus 0.69 problem with long recipient list

pgnus 0.69 problem with long recipient list

[Doug Bagley]

Someone sent me an email with a long (30 recipients) To: header.
pgnus 0.69 hangs reading this mail, but pgnus 0.66 doesn't.  I spent
some time fiddling around testing it, so I'm sure it's a real problem
and not my allergy medicine causing hallucinations.  Through a fumble
finger mistake I lost the mail, and I'm unable to reproduce the
problem just by faking up a similar one.  It must have been some
peculiar combination of the headers plus the long recipient list.  If
I can reproduce it again, I'll send a better report.

Cheers,
Doug
-- 
To continue press any key.
Homer: "Where's the any key?"

Re: pgnus 0.69 problem with long recipient list

[Doug Bagley]

[-- Attachment #1: Type: text/plain, Size: 174 bytes --]

I think it was just a long recipient list which hosed things up
... anyway, here's the example email that chokes my pgnus-0.69.  I'm
enclosing it with this cool MIME thing:


[-- Attachment #2: blurfl --]
[-- Type: text/plain, Size: 1250 bytes --]

>From foo  Mon Jan  4 19:11:00 1999
From: foo <foo@foo.com>
To: "Doug Bagley" <doug@bagley.org>,
	User Account 1 <user1@bagley.org>,
	User Account 2 <user2@bagley.org>,
	User Account 3 <user3@bagley.org>,
	User Account 4 <user4@bagley.org>,
	User Account 5 <user5@bagley.org>,
	User Account 6 <user6@bagley.org>,
	User Account 7 <user7@bagley.org>,
	User Account 8 <user8@bagley.org>,
	User Account 9 <user9@bagley.org>,
	User Account 10 <user10@bagley.org>,
	User Account 11 <user11@bagley.org>,
	User Account 12 <user12@bagley.org>,
	User Account 13 <user13@bagley.org>,
	User Account 14 <user14@bagley.org>,
	User Account 15 <user15@bagley.org>,
	User Account 16 <user16@bagley.org>,
	User Account 17 <user17@bagley.org>,
	User Account 18 <user18@bagley.org>,
	User Account 19 <user19@bagley.org>,
	User Account 20 <user20@bagley.org>,
	User Account 21 <user21@bagley.org>,
	User Account 22 <user22@bagley.org>,
	User Account 23 <user23@bagley.org>,
	User Account 24 <user24@bagley.org>,
	User Account 25 <user25@bagley.org>,
	User Account 26 <user26@bagley.org>,
	User Account 27 <user27@bagley.org>,
	User Account 28 <user28@bagley.org>,
	User Account 29 <user29@bagley.org>
Message-Id: <B199901022020.MAA18071@swan.prod.itd.earthlink.net>

foo

[-- Attachment #3: Type: text/plain, Size: 78 bytes --]


Cheers,
Doug
-- 
Excessive sneezing due to allergies can cause Dain Bramage.

Re: pgnus 0.69 problem with long recipient list

[Jack Vinson]

>>>>> "DB" == Doug Bagley <doug@dejanews.com> writes:

DB> I think it was just a long recipient list which hosed things up
DB> ... anyway, here's the example email that chokes my pgnus-0.69.  I'm
DB> enclosing it with this cool MIME thing:
[snip]

The message that Doug attached in MIME caused pgnus-0.69 to barf when it
attempted to parse that message from my server.  The problem appears to
have been Gnus thinking it had found another message while it was parsing
Doug's attachment.  Why would it do that?  Is this the same bug that is
fixed by Colin Rafferty's "[patch]?"

[checking]

Well, the patch made it possible to collect the message, but the message
itself comes in as two pieces: one from Doug and one from "foo."  The one
from Doug thinks it is a multipart message, but it only contains the text
up to the "From foo" line.

Obviously, something strange is happening when Gnus parses a multipart
message that just happens to contain another message.

-- 
Jack Vinson <jvinson@chevax.ecs.umass.edu>    http://www.cis.upenn.edu/~vinson/
Zippy: Mr and Mrs PED, can I borrow 26.7% of the RAYON TEXTILE production
 of the INDONESIAN archipelago?

Re: pgnus 0.69 problem with long recipient list

[Lars Magne Ingebrigtsen]

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

> The message that Doug attached in MIME caused pgnus-0.69 to barf when it
> attempted to parse that message from my server.  The problem appears to
> have been Gnus thinking it had found another message while it was parsing
> Doug's attachment.  Why would it do that? 

Because the delivery agent you're using probably hadn't quoted the
"From " envelope header that was included in the message.

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

Re: pgnus 0.69 problem with long recipient list

[Lars Magne Ingebrigtsen]

Doug Bagley <doug@dejanews.com> writes:

> I think it was just a long recipient list which hosed things up
> ... anyway, here's the example email that chokes my pgnus-0.69. 

I've now lowered nnmail-split-header-length-limit to 512 in 0.70.

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

Re: pgnus 0.69 problem with long recipient list

[Karl Kleinpaste]

Doug Bagley <doug@dejanews.com> writes:
> Someone sent me an email with a long (30 recipients) To: header.
> pgnus 0.69 hangs reading this mail, but pgnus 0.66 doesn't.

I had this happen to me today.  I set debug-on-quit and tried again,
and got this stack when I knocked it loose with ^G:

Signaling: (quit)
  nnmail-article-group(nnml-active-number)
  nnmail-check-duplication("<199901042238.RAA31299@security.wcom.net>" nnml-save-mail nnml-active-number)
  nnmail-process-unix-mail-format(nnml-save-mail nnml-active-number)
  nnmail-split-incoming("/home/karl/Mail/.private/.gnus-crash-box" nnml-save-mail nil "personal.jprc" nnml-active-number)
  nnmail-get-new-mail(nnml nnml-save-nov "~/Mail/" "personal.jprc")
  nnml-request-scan("personal.jprc" "")
  gnus-request-scan("nnml:personal.jprc" (nnml ""))
  gnus-activate-group("nnml:personal.jprc" scan)
  gnus-get-unread-articles(2)
  gnus-group-get-new-news(2)
  call-interactively(gnus-group-get-new-news)

The problem is consistently reproducible.  The To: list is 27 lines
and 1818 chars.  (All legit, not spam -- it's a friend informing a
mess of people that he's changing jobs.)

Re: pgnus 0.69 problem with long recipient list

[Colin Rafferty]

Doug Bagley writes:

> Someone sent me an email with a long (30 recipients) To: header.
> pgnus 0.69 hangs reading this mail, but pgnus 0.66 doesn't.

See my message from yesterday with the subject "[patch]".

Sorry about the not-quite-useful subject, but I forgot.

-- 
Colin