>Thanks, Bill. Some very useful info there. You're welcome. :-) Thanks for your help as well. >So, it appears that trn was told that the group had a low-article of 701 >(presumably by the result of the GROUP command), but the LISTGROUP >command returned an article 700. I'd be interested in knowing if this >is a consistent off-by-one error in that news server software, or if the >error can vary from group to group (i.e. is it ever greater than 1?). I recompiled bits.c with optimization turned off so I could put breakpoints in which would let me examine the 'an' value with gdb. I then spent some time 'reading' random newsgroups off of the NNTP server trying to trigger the problem. I found my conditional breakpoint being hit more often then I would have guessed based on my frequency of coredumps. When I continued, I didn't always get a SEGV, instead trn would often present the group to me as being empty. I found that usually it was off-by-one, but at least one time it was off-by-two. (This is seeing the problem maybe 3-5 times on Friday evening.) If I exited trn, modified the newsrc to let me re-read the group, and restarted trn; I found that I did not see the problem in that newsgroup again. Bill Bogstad bogstad@pobox.com P.S. I haven't actually tried any of the patches yet. The next thing I'll probably do is see if the articles are actually there. If they are I'ld prefer not to have trn ignore them... ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf
