differences between fread and read

[CDiMa]

Hi!

When I have to deal with big inputs I usually deal with it reading in a buffer the data:
[c]numread=fread(buffer, sizeof(char), MAXBUF, stdin);[/c]
or:
[c]numread=read(0, buffer, MAXBUF);[/c]
What are the differences between these two?
On my system it looks like read doesn't try to fill the whole buffer, but only gets as far as the first newline .
Does this have to do with some settings of the input stream? May I rely on this behaviour?
Is there a way to explicitly enable/disable it?

Thanks in advance for any suggestion!

Ciao!!!

Claudio

[Dominik Michniewski]

fread() does buffered read, but read() - unbuffered. So read() is called "lower API function" - fread() uses read() and it's only a fascade (I think)

Best reagrds
DM

[Ivor]

I'm not going to check it but as I remember read() uses buffer too. There's some sort of constant or setting in lower functions that enables you to control buffering. There's three choices: fully buffered, line buffered and not buffered. Because read() is a system function it can vary a bit in it's declaration (see borland c earlier versions, if I'm correct even declaration of fread() was a bit different, don't remember though, 's been some time since I used that one). Because of that it is advisable to use fread() -- it's ANSI.

I'm not giving my head for the things I wrote.
Ivor

[CDiMa]

There's some sort of constant or setting in lower functions that enables you to control buffering. There's three choices: fully buffered, line buffered and not buffered.

Anyone can say more on this? How can one control how read buffers data?

Ciao!!!

Claudio

[Russell John]

There are two ways to read in C - fscan and fread... read is like fread in C, to read byte by byte.

[c]inp.read(char *buffer, int n);[/c]
Char *buffer, where to read to, n is number of bytes to read.

[CDiMa]

There are two ways to read in C - fscan and fread... read is like fread in C, to read byte by byte.

[c]inp.read(char *buffer, int n);[/c]
Char *buffer, where to read to, n is number of bytes to read.

Hmmm, I'm not sure if we are talking of the same thing...
On a UNIX system read is a system call to read data from a stream.
In C read is a function provided by the standard C library that takes a file descriptor and uses the read system call to read data.
I don't know what inp.read means, I guess you are talking about C++.

Anyway my problem is very specific.
On my system when I try to read() say 1024 bytes I always get only a few bytes up to the first newline. This suggests that some kind of buffering takes place at a lower level than the C read() function. I'd like to know if there is some way to enable/disable this behaviour.

I hope this is a little bit clearer now...

Ciao!!!

Claudio