Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!uwvax!zazen!decwrl!pa.dec.com!decprl!decprl!boyd
From: boyd@prl.dec.com (Boyd Roberts)
Newsgroups: comp.unix.questions
Subject: Re: Can't cat tape- big blocks?
Message-ID: <1991Jun19.093448.23695@prl.dec.com>
Date: 19 Jun 91 09:34:48 GMT
References: <803@adpplz.UUCP> <1991Jun14.094822.7029@prl.dec.com> <14433@dog.ee.lbl.gov>
Sender: news@prl.dec.com (USENET News System)
Reply-To: boyd@prl.dec.com (Boyd Roberts)
Distribution: usa
Organization: Digital Equipment Corporation - Paris Research Laboratory
Lines: 17
Nntp-Posting-Host: prl313.prl.dec.com

In article <14433@dog.ee.lbl.gov>, torek@elf.ee.lbl.gov (Chris Torek) writes:
> It seems to me that the tape driver should return an error if you
> ask for 1K and the tape drive reads 10K.  Unfortunately, there is
> no obvious errno for this (ENOMEM? EINVAL? E2BIG? EFBIG? EMSGSIZE?
> ENOBUFS?).

I saw one driver hacked to return the amount not read.  No, not one
of my hacks.  I'm not sure whether it was such a good idea though.

Programs who blindly believe write() > 0 is ok, just won't work.
Smart archivers could benefit from it, but I think the cost
of broken programs would be too high.


Boyd Roberts			boyd@prl.dec.com

``When the going gets wierd, the weird turn pro...''