flee@shire.cs.psu.edu (Felix Lee) (10/29/89)
I wrote: > Yes, except you need a "maybe-I-have" NNTP response, if you want > reliable article transmission. John Coolidge <coolidge@brutus.cs.uiuc.edu> wrote: > It's in the NNTP spec, more or less. Just use NNTP response code 436 > (transfer failed - try again later). Except that when "nntpxmit" gets a "436" it aborts the rest of the transfer. "436" is used by "nntpd" when "rnews" fails, and when that happens, it's often pointless to keep sending things to have "rnews" fail again. Some distinction should be made between this-article failure and news-system failure. -- Felix Lee flee@shire.cs.psu.edu *!psuvax1!flee
flee@shire.cs.psu.edu (Felix Lee) (10/29/89)
Addendum, the behavior I described, "nntpxmit" aborting transmission when it gets a 436 XFERFAIL response, looks like it appeared with NNTP 1.5.4 (May 1989), along with the diskspace code in "nntpd", since 436 is also used for out-of-diskspace. -- Felix Lee flee@shire.cs.psu.edu *!psuvax1!flee
coolidge@brutus.cs.uiuc.edu (John Coolidge) (10/30/89)
flee@shire.cs.psu.edu (Felix Lee) writes: >John Coolidge <coolidge@brutus.cs.uiuc.edu> wrote: >> It's in the NNTP spec, more or less. Just use NNTP response code 436 >> (transfer failed - try again later). >Except that when "nntpxmit" gets a "436" it aborts the rest of the >transfer. "436" is used by "nntpd" when "rnews" fails, and when that >happens, it's often pointless to keep sending things to have "rnews" >fail again. ACK! Well, I _said_ I wasn't sure that current software handled it correctly! :-) >Some distinction should be made between this-article failure and >news-system failure. You're right, of course. We seem to need a new response code for NNTP to handle this, since the existing one has been used to deal with a different class of errors. --John -------------------------------------------------------------------------- John L. Coolidge Internet:coolidge@cs.uiuc.edu UUCP:uiucdcs!coolidge Of course I don't speak for the U of I (or anyone else except myself) Copyright 1989 John L. Coolidge. Copying allowed if (and only if) attributed. You may redistribute this article if and only if your recipients may as well.
brian@ucsd.Edu (Brian Kantor) (10/30/89)
The 436 response code is intended to signify a failure to post the instant article. It does NOT imply complete news system failure; a 5xx code should be used for that. If you have a temporary failure (such as running out of disk space) and someone attempts to send you an article, you can 436 it (assuming that he'll send it again later) and then continue, or if you're going to give up, 400 it and close the connection. I thought the current version of nntpxmit requeued articles that were 4xx'd. Maybe I read the code wrong. - Brian