Re: tin ends abruptly

Christian Knoke wrote:
> On Sun, Mar 28, 2004 at 04:14:20PM +0200, Dirk Nimmich wrote:
> 
> > > 200 Leafnode NNTP Daemon, version 1.9.19 running at
> > > bolte.at.home
> > 
> > Before doing anything else, please upgrade this ancient version
> > of Leafnode. It has numerous bugs, including security related
> > ones.

I forgot to mention bugs that cause data loss.

> Besides, that it was running for several years now without any
> trouble, you are probably right.

That was probably only by accident. :-) Just as a side note, the
author of leafnode refuses to support versions of leafnode older
than 1.9.50. 1.9.19 in particular must have been a pretty bad
release, unfortunately it is very wide spread because several
distributions still include it in their "stable" versions.

I remember several postings in de.comm.software.newsserver where
Matthias was steadily encouraging people to update, and his policy
to don't support older versions is a direct result of the lazyness
of these people and the distribution maintainers who refused to do
so. He was tired of repeating over and over again that most of their
problems would vanish immediately if they only upgraded to the
latest version.

> > > Mar 28 14:48:14 bolte leafnode[5571]: <XOVER 2467-2766
> > > Mar 28 14:48:14 bolte leafnode[5571]: >224 Overview information for postings 2467-2766:
> > > Mar 28 14:48:14 bolte leafnode[5571]: >400 Service discontinued
> >                                          ^^^^^^^^^^^^^^^^^^^^^^^^
> > 
> > This is the main offense. This is an unexpected response, and tins
> > reaction is to stop running (continue to run wouldn't even help in
> > this case). Servers should never send a response code without being
> > asked to do anything.
> 
> Ok, then. Maybe tin should print a notice here: unexpected response from
> server?

Well, maybe yes. ;-) The problem is "here" because there are a lot
of "here"s in the code (in fact, everywhere where tin sends a
command and interprets the response code) and sometimes it's
possible to continue while at another time it's not. I'm not even
sure if there are already code paths that cover some of these cases
and handle it gracefully.

Patches are welcome, even if it only affects doc/TODO. ;-) Please
use tin-dev if you want to discuss this topic any further.