continuing problem with gnus-change-server

continuing problem with gnus-change-server

[Dan Christensen]

I've been putting off changing nntp servers for a few months, but now I've
got only 7 days before I lose access to old news server.  However,
gnus-change-server is still giving me problems, with qgnus 0.14 and
emacs 19.34.1.  This has been discussed before, but below there's
something that might provide a hint.  Here's a backtrace:

Signaling: (wrong-type-argument number-or-marker-p nil)
  <(nil nil)
  sort((nil) <)
  gnus-move-group-to-server(("sci.math" 4 ((1 . 210029)) ((cache 171741 173689 199399 199778) (tick 171741 173689 199399 199778))) (nntp "mit" (nntp-address "runge.mit.edu") (nntp-open-connection-function nntp-open-rlogin) (nntp-end-of-line "\n") (nntp-rlogin-user-name "jdchrist") (nntp-rlogin-parameters ("telnet" "galois.mit.edu" "nntp"))) (nntp "news.jhu.edu"))
  gnus-change-server((nntp "mit" (nntp-address "runge.mit.edu") (nntp-open-connection-function nntp-open-rlogin) (nntp-end-of-line "\n") (nntp-rlogin-user-name "jdchrist") (nntp-rlogin-parameters ("telnet" "galois.mit.edu" "nntp"))) (nntp "news.jhu.edu"))
* call-interactively(gnus-change-server)
  execute-extended-command(nil)
  call-interactively(execute-extended-command)

If I execute (sort '(nil) '<), emacs successfully returns (nil).
However, if I execute (sort '(nil nil) '<), emacs complains and gives
me the following backtrace, which looks suspiciously like the above
backtrace.

Signaling: (wrong-type-argument number-or-marker-p nil)
  <(nil nil)
  sort((nil) <)
  eval((sort (quote (nil)) (quote <)))
  eval-expression((sort (quote (nil)) (quote <)))
  call-interactively(eval-expression)

Could gnus be trying to sort a list like (nil nil)?  (And can anyone
explain why in the backtrace it says sort((nil) <) rather than
sort((nil nil) <)?)

Dan

Re: continuing problem with gnus-change-server

[Lars Magne Ingebrigtsen]

jdc@chow.mat.jhu.edu (Dan Christensen) writes:

> Signaling: (wrong-type-argument number-or-marker-p nil)
>   <(nil nil)
>   sort((nil) <)

[...]

> If I execute (sort '(nil) '<), emacs successfully returns (nil).

There is something odd about sort.  It sometimes works, and sometimes
it doesn't.

> Could gnus be trying to sort a list like (nil nil)?

No.

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

Re: continuing problem with gnus-change-server

[Felix Lee]

> > Signaling: (wrong-type-argument number-or-marker-p nil)
> >   <(nil nil)
> >   sort((nil) <)
> > If I execute (sort '(nil) '<), emacs successfully returns (nil).
> There is something odd about sort.  It sometimes works, and sometimes
> it doesn't.
> > Could gnus be trying to sort a list like (nil nil)?
> No.

hmm.  that's the only way I can get this failure in
emacs-20.2.  the backtrace is a little misleading though.

    (sort '(nil nil nil nil nil) '<)

    Signaling: (wrong-type-argument number-or-marker-p nil)
      <(nil nil)
      sort((nil) <)
      eval((sort (quote (nil)) (quote <)))
--

Re: continuing problem with gnus-change-server

[Hrvoje Niksic]

Felix Lee <flee@teleport.com> writes:

> hmm.  that's the only way I can get this failure in
> emacs-20.2.  the backtrace is a little misleading though.

It all makes sense now.  The backtrace is FUBAR because `sort' changes 
its first argument destructively!

-- 
Hrvoje Niksic <hniksic@srce.hr> | Student at FER Zagreb, Croatia
--------------------------------+--------------------------------
ED WILL NOT CORRUPT YOUR PRECIOUS BODILY FLUIDS!!

Re: continuing problem with gnus-change-server

[Lars Magne Ingebrigtsen]

Hrvoje Niksic <hniksic@srce.hr> writes:

> > hmm.  that's the only way I can get this failure in
> > emacs-20.2.  the backtrace is a little misleading though.
> 
> It all makes sense now.  The backtrace is FUBAR because `sort' changes 
> its first argument destructively!

Ah!  It all makes sense now.

One more of life's big mysteries solved.  I'll slap some `(delq nil ...)'
around some lists that I feed to sort, and Bob's your uncle.
(Whatever that means.)

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

Re: continuing problem with gnus-change-server

[Darren/Torin/Who Ever...]

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

Lars Magne Ingebrigtsen, in an immanent manifestation of deity, wrote:
>One more of life's big mysteries solved.  I'll slap some `(delq nil ...)'
>around some lists that I feed to sort, and Bob's your uncle.
>(Whatever that means.)

Well, from a friend of mine that collects such things:
>It comes out of the history of England. Some time ago (a century or
>two?) there was a guy who got a political appointment because his uncle
>was the rt. hon. Robert somethingorother (someone famous, but not to us
>here, and I forget his name). Hence the idea that if you want some
>particular political plum, you can get it if Bob's your uncle... and it
>spreads from there. Sorta similar to "Q. E. D." and "that's all she
>wrote."

He used it when I was explaining pgp to him and I asked him.
Unsurprisingly he knew.  That's a tech editor for you.

Darren
- -- 
<torin@daft.com> <http://www.daft.com/~torin> <torin@debian.org> <torin@io.com>
Darren Stalder/2608 Second Ave, @282/Seattle, WA 98121-1212/USA/+1-800-921-4996
@ Sysadmin, webweaver, postmaster for hire.  C/Perl/CGI programmer and tutor. @
@		     Make a little hot-tub in your soul.		      @

-----BEGIN PGP SIGNATURE-----
Version: 2.6.3
Charset: noconv
Comment: Processed by Mailcrypt 3.4, an Emacs/PGP interface

iQCVAwUBNH/xAo4wrq++1Ls5AQHTGAP/QfyoO57hJ99oAcaE6rgtlwgN8GJhBijJ
zyUCe4WyD2ZmMUbIkthNSmqXFqFyEevNJeo75pnjhf9R9ZZpikgSHt70ZEQ/eE/u
GkyYKLx218FgAe6A5fJkQhD+vG2A7s17762L7CmkM4inb+QT9Nz5hGN51OCy6Ydd
RcEqOPH5e30=
=lh6W
-----END PGP SIGNATURE-----

Re: continuing problem with gnus-change-server

[Michael Welsh Duggan]

"Darren/Torin/Who Ever..." <torin@daft.com> writes:

> -----BEGIN PGP SIGNED MESSAGE-----
> 
> Lars Magne Ingebrigtsen, in an immanent manifestation of deity, wrote:
> >One more of life's big mysteries solved.  I'll slap some `(delq nil ...)'
> >around some lists that I feed to sort, and Bob's your uncle.
> >(Whatever that means.)
> 
> Well, from a friend of mine that collects such things:
> >It comes out of the history of England. Some time ago (a century or
> >two?) there was a guy who got a political appointment because his uncle
> >was the rt. hon. Robert somethingorother (someone famous, but not to us
> >here, and I forget his name). Hence the idea that if you want some
> >particular political plum, you can get it if Bob's your uncle... and it
> >spreads from there. Sorta similar to "Q. E. D." and "that's all she
> >wrote."

Yup.  Definitely correct.

<URL:http://www.word-detective.com/back-f.html#bob>

-- 
Michael Duggan
(md5i@cs.cmu.edu)

That `sort' bug...

[Lars Magne Ingebrigtsen]

Signaling: (wrong-type-argument number-or-marker-p nil)
  <(nil nil)
  sort((nil) <)

Someone discovered why this happened, but I've forgotten what the
solution was.  Could that someone (or someone else) remind me?  Pretty
please?

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

Re: That `sort' bug...

[SL Baur]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Signaling: (wrong-type-argument number-or-marker-p nil)
>   <(nil nil)
>   sort((nil) <)

> Someone discovered why this happened, but I've forgotten what the
> solution was.  Could that someone (or someone else) remind me?  Pretty
> please?

Well, since you asked nicely ... :-)

My records show it was an XEmacs 19.15 bug not present in XEmacs 20.
If you have archives of this list back to late May, early June 1997,
the thread was entitled `gnus-change-server problems?' and the root
Message-ID was <wpn2pczsig.fsf@CompleteIS.com>.

Re: That `sort' bug...

[Hrvoje Niksic]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Signaling: (wrong-type-argument number-or-marker-p nil)
>   <(nil nil)
>   sort((nil) <)
> 
> Someone discovered why this happened, but I've forgotten what the
> solution was.  Could that someone (or someone else) remind me?

`sort' modifies its first argument destructively, thus invalidating
the backtrace.  This generates the same kind of backtrace:

    (sort '(nil nil nil nil) '<)

The solution was to add `(delq nil LIST)' in an appropriate place or
two.

-- 
Hrvoje Niksic <hniksic@srce.hr> | Student at FER Zagreb, Croatia
--------------------------------+--------------------------------
Unspeakable horrors from outer space paralyze the living and
resurrect the dead!

Re: That `sort' bug...

[Lars Magne Ingebrigtsen]

Hrvoje Niksic <hniksic@srce.hr> writes:

> `sort' modifies its first argument destructively, thus invalidating
> the backtrace.  This generates the same kind of backtrace:
> 
>     (sort '(nil nil nil nil) '<)
> 
> The solution was to add `(delq nil LIST)' in an appropriate place or
> two.

Ah; right.  I've now added it one more place.  :-)

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

Re: That `sort' bug...

[Dan Christensen]

The problem is that sort is being told to sort a list containing nil
elements using <, but < can't compare nil elements.  The reason that
(sort '(nil) '<) doesn't produce an error is that no comparisons need
to be done.  However, (sort '(nil nil) '<) does produce an error.

The backtrace for the error gnus gets shows sort((nil) <).  However,
since sort modifies its arguments, this is misleading.  The backtrace
for the error produced by (sort '(nil nil) '<) also shows sort((nil) <).

Attached below are the relevant articles, trimmed a bit.

Dan

jdc@chow.mat.jhu.edu (Dan Christensen) writes:
> gnus-change-server is still giving me problems, with qgnus 0.14 and
> emacs 19.34.1.  
> 
> Signaling: (wrong-type-argument number-or-marker-p nil)
>   <(nil nil)
>   sort((nil) <)
[...]
> * call-interactively(gnus-change-server)
[...]
> 
> If I execute (sort '(nil) '<), emacs successfully returns (nil).
> However, if I execute (sort '(nil nil) '<), emacs complains and gives
> me the following backtrace, which looks suspiciously like the above
> backtrace.
> 
> Signaling: (wrong-type-argument number-or-marker-p nil)
>   <(nil nil)
>   sort((nil) <)
>   eval((sort (quote (nil)) (quote <)))
>   eval-expression((sort (quote (nil)) (quote <)))
>   call-interactively(eval-expression)
> 
> Could gnus be trying to sort a list like (nil nil)?  (And can anyone
> explain why in the backtrace it says sort((nil) <) rather than
> sort((nil nil) <)?)

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:
> > If I execute (sort '(nil) '<), emacs successfully returns (nil).
> 
> There is something odd about sort.  It sometimes works, and sometimes
> it doesn't.
> 
> > Could gnus be trying to sort a list like (nil nil)?
> 
> No.

[I don't see any evidence that there is a problem with sort.  It works
when it doesn't need to compare incomparable elements.]

Hrvoje Niksic <hniksic@srce.hr> writes:
> Felix Lee <flee@teleport.com> writes:
> 
> > hmm.  that's the only way I can get this failure in
> > emacs-20.2.  the backtrace is a little misleading though.
> 
> It all makes sense now.  The backtrace is FUBAR because `sort' changes 
> its first argument destructively!

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:
> 
> Ah!  It all makes sense now.
> 
> One more of life's big mysteries solved.  I'll slap some `(delq nil ...)'
> around some lists that I feed to sort, and Bob's your uncle.
> (Whatever that means.)

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:
> Signaling: (wrong-type-argument number-or-marker-p nil)
>   <(nil nil)
>   sort((nil) <)
> 
> Someone discovered why this happened, but I've forgotten what the
> solution was.  Could that someone (or someone else) remind me?  Pretty
> please?