Always 224 with XOVER?

[Julien ÉLIE]

Hi,

nnrpd always answers

224
.

when there is nothing to send for XOVER.

GROUP fr.test
211 35 80679 80716 fr.test
XOVER 1
224 1 fields follow
.

Yet, "if no articles are in the range specified, a 420 error response is
returned by the server" according to RFC 2980 for XOVER...

Too bad that INN, Diablo and Dnews (the three I checked by telnet'ing
news servers) send 224 for that case.
By the way, it is amusing to telnet news servers.  I saw that the one
of my ISP (Orange) is an old version of INN which implements SLAVE
and the distributions field to NEWNEWS and NEWGROUPS but they removed
the mention of INN and they added support for HDR and OVER (but when
I tested, I saw that they were only aliases for XHDR and XOVER!).
Such changes will not help to implement standards!


Well, the reason why I write is that I changed 224 to 420 when needed
for XOVER, thinking that it would be right.  Unfortunately, some
news clients (like Windows Mail...) send a pop-up to tell me
"Error 420:  No articles in 976-976"!!!!

Damn it!

nnrpd -t tells me:

200 news.trigofacile.com InterNetNews NNRP server INN 2.5.0 (20080629 prerelease) ready (no posting)
MODE READER
200 news.trigofacile.com InterNetNews NNRP server INN 2.5.0 (20080629 prerelease) ready (no posting)
GROUP fr.usenet.distribution
211 649 1 976 fr.usenet.distribution
XOVER 976-976
420 No articles in 976-976

As a matter of fact, the last article I had was 975 and the 976th was cancelled!


I tried with Thunderbird and it does not trigger any error off.
I do not know however why it tries HEAD -- maybe it does not rely on the result
of XOVER :)

200 news.trigofacile.com InterNetNews NNRP server INN 2.5.0 (20080629 prerelease) ready (no posting)
MODE READER
200 news.trigofacile.com InterNetNews NNRP server INN 2.5.0 (20080629 prerelease) ready (no posting)
GROUP fr.usenet.distribution
211 649 1 976 fr.usenet.distribution
XOVER 976-976
420 No articles in 976-976
HEAD 976
423 Bad article number



A few news servers correctly implement XOVER (for instance the Columbus nntp.perl.org)
but I believe we should not try to "improve" XOVER.

Russ, you're right:  it's a bad idea to fix wrong implementations :(
However, I wonder how people who want to implement XOVER will find out that they should
not return 420...

Well, I will have to rewrite OVER/XOVER to handle yet another difference...

-- 
Julien ÉLIE

« Tout homme devrait un jour se marier ;
  après tout, le bonheur n'est pas la seule chose dans la vie. »