NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/std/c/10

Internet Message Format | 1996-08-06 | 1.9 KB

Path: lyra.csx.cam.ac.uk!nmm1
From: [e-mail] (Nick Maclaren)
Newsgroups: comp.std.c
Subject: Re: Buffering??  What buffering??
Date: 2 Jan 1996 10:43:33 GMT
Organization: University of Cambridge, England
Message-ID: <4cb28l$[e-mail]>
References: <4c9uk4$[e-mail]>
NNTP-Posting-Host: ursa.cus.cam.ac.uk

In article <4c9uk4$[e-mail]>, [e-mail] (Ronald F. Guilmette) writes:
|> I would like to know the exact semantics of the _IOFBF, _IOLBF, and
|> _IONBF mode flags for the `setvbuf' function (subclause 7.9.5.6 of the
|> ANSI/ISO C tandard).
|>
|> Correct me if I'm wrong, but there doesn't seem to be any description
|> whatsoever of the semantics of these flags in the C standard.  Given
|> that, I rather wonder why these were standardized at all.  Certainly,
|> to just say ``Here are some possible flags for this function.  We don't
|> know what they do.'' doesn't help anybody.

See section 4.9.3 of the ANSI standard or the equivalent in the ISO one,
but here is a rough description:

_IONBF causes each character to be read or written as it happens.

_IOLBF reads or writes data when the buffer is full or when a logical
newline is encountered (and therein hangs a tale!)

_IOFBF does this only when the buffer is full.

fflush and fclose (but not necessarily rewind) cause the reading or
writing to take place immediately.


But remember that reading and writing refer only to C's interface
with its environment, and not necessarily to the interaction of that
environment with other environments or peripherals.  Until you have
tried to do it (and probably failed), you cannot imagine how foul it
is to write a message synchronously even within a single UNIX system.
And as for networks :-(


Nick Maclaren,
University of Cambridge Computer Laboratory,
New Museums Site, Pembroke Street, Cambridge CB2 3QG, England.
Email:  [e-mail]
Tel.:  [phone]    Fax:  [phone]