Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.1 6/24/83; site cincy.UUCP Path: utzoo!watmath!clyde!burl!ulysses!mhuxl!cbosgd!qusavx!cincy!jreuter From: jreuter@cincy.UUCP (Jim Reuter) Newsgroups: net.unix Subject: EOF character changing Message-ID: <1173@cincy.UUCP> Date: Tue, 14-Feb-84 00:03:40 EST Article-I.D.: cincy.1173 Posted: Tue Feb 14 00:03:40 1984 Date-Received: Fri, 17-Feb-84 02:31:52 EST References: <964@proper.UUCP>, <54400001@hpfcra.UUCP> Organization: U. of Cincinnati ECE Lines: 18 The reason System V (and System III and System whatever-else-Bell-sells) changes the EOF character is because of the raw-mode delays they added. When in raw mode in v7, every character coming in wakes up the process reading the line, which can swamp a system. System III and friends in raw mode have both a minimum-character threshold and a time threshold before waking up the reading process, the intent being a reduction of system load. Unfortunately, they decided to multiplex the character-count and time thresholds into already existing variables, namely the EOF and the BRK (I think) character. So, the ioctl code automatically sets these fields to the 'appropriate' values when switching between raw and cooked mode (ICANON mode for those systems). Assuming, of course, that you consider their canned values 'appropriate'. Don't hold me to the constant and variable names mentioned above, I don't have the stuff in front of me and it has been a while. Jim