K 10
svn:author
V 5
eagle
K 8
svn:date
V 27
2005-10-09T04:36:11.008843Z
K 7
svn:log
V 322
Fix an oddity that I found a while back when auditing uses of the buffer
code in innd but never returned to.  If NCwritereply sends its reply
immediately but can only send a part of it, the count of bytes remaining
to send wasn't decremented.  We could then send random junk at the end
of the buffer to the peer later on.

END
