Problems using download2day

[zx81]

Hi,
I tried to move from giganews to download2day.

The problem is that urd can't subscribe to newsgroups. (I cleaned the whole env and used the svn-version from today).

It fails in the guess_first_article() with an: Base_NNTP_Client->cmdNext() giving "Base_NNTP_Client->_handleUnexpectedResponse(0)".

Any clues what might be wrong ? Credentials are OK, i can login and get the newsgroups-list.

TIA,
-Markus

[spearhead]

Hi,
I tried to move from giganews to download2day.

The problem is that urd can't subscribe to newsgroups. (I cleaned the whole env and used the svn-version from today).

It fails in the guess_first_article() with an: Base_NNTP_Client->cmdNext() giving "Base_NNTP_Client->_handleUnexpectedResponse(0)".

Any clues what might be wrong ? Credentials are OK, i can login and get the newsgroups-list.

TIA,
-Markus

Not really...but I noticed that the newsserver gives a different response when not logged in to the server on the "next" command then is currently in the code. A added that response explicedly. Maybe it does that too for other commands. I also added some more error handling.  make sure that your log level is at least Notice (or info or debug) in admin/configuration to see these messages. Can you post the log trace ?

[zx81]

Hi,
I just tried the svn-version but i made no difference.

From my understanding of the code the guess_first_article() simply issues a "group" followed by "next" - correct ?

I tried that via netcat and this is the result:
group alt.binaries.boneless
211 1717732060 683169 1718415228 alt.binaries.boneless
next
421 No next to retrieve

Note that after issuing the "next" it took the server about 4 minutes to come back with the 421.

So i my case the function my actually fail due to a timeout.
So question: where to set the timeout in the nntp-class (just to track it down). Also: can guess_first_arcticle handle the "no next to retrieve" ? I suppose the function is designed to expect the first valid article-id here.

Thanks,
-Markus

[spearhead]

what is does is trying to skip articles that would expire anyway and do that in a sort of binary search method, so that the number articles to download is minimised.  For a huge group like boneless with a retention of 100 days, downloading would take forever... and if the expire  is set to 5 days, most of it would have to be discarded anyway.

the next should yield something like

Code:

223 66567098 <part43of235.tnIE5RqOqsAHbnBZpid&@youwouldnt-download.a.car.local> Article retrieved; request text separately.

apparently either the ng server is broken cq does not conform to the rfc. It's answer suggests that the ng is empty (which is unlikely in the case of boneless). No next actually sets an internal pointer on the server to the first valid article after the group command. The four minutes is not a good sign either. what does the LAST command give after a GROUP command?

I improved the timeout functions a bit (timeout is 15 secs by default) and made some cleanup in  guess...() code which is still a bit messy ... but it should be more resilient to erros.

[spearhead]

Does giganews allow indexing ?

[zx81]

Yes. I used that provider without any problems.

-Markus