Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site hocda.UUCP
Path: utzoo!linus!decvax!harpo!eagle!mhuxl!houxm!hocda!54394gt
From: 54394gt@hocda.UUCP (G.TOMASEVICH)
Newsgroups: net.unix
Subject: Raw vs Block Tape I/O
Message-ID: <341@hocda.UUCP>
Date: Thu, 12-Jan-84 12:55:57 EST
Article-I.D.: hocda.341
Posted: Thu Jan 12 12:55:57 1984
Date-Received: Fri, 13-Jan-84 07:04:19 EST
Organization: Bell Labs, Holmdel
Lines: 8

There was a comment on rarely using the block version of mag-tape i/o.  On our
system, USG 4.2, 'cat' does not flush the last block if only partially full,
for block tape writes, so block i/o is worse than useless in that case.
Many commands seem to work either way on tape, though.  I prefer using raw, or
character-special mode, for reading, since it will correctly read any size
tape block that fits into the supplied buffer.  The main advantage of raw
tape i/o is using large buffers (and hence record size) to minimize the
amount of tape used up for inter-record gaps.