mp@mit-eddi.UUCP (06/03/83)
For the past 2 days, we've been getting quite a few (10 per day) "Inbound news is garbled" errors. We (and mit-vax) run 2.9, and our neighbors run 2.10. I tracked the problem down to 2.9's hread and frmread routines (in header.c), which will return NULL if they don't find a "From" (or "Path"), "Posted", and "Article-I.D." line in the message's header. Each offending message contained a "Message-ID" line in the header, but no "Article-I.D." line! The messages originate from various machines; possibly some machine in the interim is stripping off the "Article-I.D." line. I guess the easiest thing to do is include one offending message and ask all the machines along the path to check if they indeed have an "Article-I.D." line in the header of their copy of the message. Here it is. [this message is from mit-vax, not mit-eddie, by the way] --------------------------------------------------------------------------------- |Relay-Version: version B 2.10 beta 3/9/83; site grkermit.UUCP |Path: grkermit!genrad!linus!allegra!eagle!harpo!seismo!hao!hplabs!sri-unix!@brl:HNIJ%mit-oz@BRL |From: %brl:HNIJ%mit-oz@BRL@sri-unix.UUCP |Newsgroups: net.micro |Subject: C for 6502 |Message-ID: <1634@sri-arpa.UUCP> |Date: Mon, 30-May-83 04:06:00 EDT |Lines: 13 | |From: John S. Labovitz <@brl:HNIJ%mit-oz@BRL> | |Does anyone know of a version of C for the 6502? I remember hearing |about Small-C ported to the 6502, but I don't know where or when. |This is to be used on an Ohio Scientific computer, but I will accept |anything for any 6502 machine. | | Thank much, | | John Labovitz, | | HNIJ%OZ @ MIT-MC |------- --------------------------------------------------------------------------------- The reason we don't run 2.10 yet is because I'm waiting for it to stabilize. Nobody here has the time to go installing every proposed fix that comes along on the net. It would be greatly appreciated if a "bug-free" version were placed on decvax and ucbvax in a week or so. Mark
chris@grkermit.UUCP (06/03/83)
I am the culprit. I knew about the OLD flag, but never checked to make sure that it was on in my defs.h. I had assumed that it was on by default when the system came out of the tar file. I am recompiling now.
mp@mit-eddi.UUCP (06/03/83)
Even after Chris fixed his source, I still found some articles being rejected as garbled. It turns out that if any header line is longer than BUFLEN chars, the article is considered garbled. I increased BUFLEN from 128 to 512. You may want to modify xxit() as I did to stash INFILE somewhere (e.g. /tmp) for later perusal if the status code is nonzero. Mark
mark@cbosgd.UUCP (06/04/83)
Article-I.D. and Posted are special headers to 2.10 - even if some site strips them out, 2.10 will (internally) put them back in. However, 2.10 only writes them out if OLD is defined. Perhaps your neighbors have OLD turned off, thus assuming that all their neighbors run 2.10 or newer?