p0.42 can't `A T'?

p0.42 can't `A T'?

[Karl Kleinpaste]

Pterodactyl Gnus v0.42
XEmacs 20.4 "Emerald" [Lucid] (i386-pc-linux) of Sun May 31 1998 on threepwood.idb.hist.no
200 news.jprc.com InterNetNews NNRP server INN 1.7.2 08-Dec-1997 ready (posting ok).

While sitting in an ordinary NNTP group served by my primary NNTP
select method, I wanted to see what preceded a given article.  `A T'
rewarded me with this stack trace:

Signaling: (error "Can't fetch thread from backends that don't support NOV")
  signal(error ("Can't fetch thread from backends that don't support NOV"))
  cerror("Can't fetch thread from backends that don't support NOV")
  apply(cerror "Can't fetch thread from backends that don't support NOV")
  error("Can't fetch thread from backends that don't support NOV")
  gnus-summary-refer-thread(nil)
  call-interactively(gnus-summary-refer-thread)

Last time I checked, normal NNTP from most any INN server supported NOV. :-)

It seems that...
	      (gnus-retrieve-headers
	       (list gnus-newsgroup-end) gnus-newsgroup-name limit)
...is returning 'headers instead of 'nov for some reason.

(In gnus-sum.el:6528.)

Re: p0.42 can't `A T'?

[Karl Kleinpaste]

I wrote:
> Signaling: (error "Can't fetch thread from backends that don't support NOV")

I set out to chase this error down.  What I found was that
nntp-server-xover was nil.  I didn't understand why that would be the
case, so I ended that Gnus session, which was a day or so old, exited
XEmacs entirely, and started again, intending to find out when
nntp-server-xover was nilified.

The error doesn't occur in the fresh XEmacs at all.

/mutter/  irreproducible bugs  /mumble/

Are there circumstances in which XOVER can fail even though the server
accepts the command?  I can only imagine that, at some point when Gnus 
sent an XOVER, though INN would have accepted it, nonetheless INN must 
have returned some complaint which Gnus (mis-?)interpreted as an
complete XOVER failure.

Re: p0.42 can't `A T'?

[Lars Magne Ingebrigtsen]

Karl Kleinpaste <karl@jprc.com> writes:

> Are there circumstances in which XOVER can fail even though the server
> accepts the command? 

Yes.  In some obscure cases -- if the first group you enter has had
all its new answers cancelled -- Gnus will believe that the nntp
server lied when it claimed to be able to grok XOVER.  (Some servers
lie on this point.)  If you know that your server really is an
XOVER-capable server, you should set `nntp-server-xover' to "XOVER"
(instead of the default `try').

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