On 13 Jan 97  10:14:10 Bob Lawrence typed to Niels Petersen ....

Hi Bob

 >> How do they read from a file in a block?

 >  NP> A file read (at the lowest level) places one sector (512 bytes)
 >  NP> into a particular place in memory allocated for that purpose.

 >   This is what Paul said, but it doesn't work in practice.

 What I wrote is _exactly_ what happens at the machine level regardless
 of what programming language you are using.

In your following statement you confirm that a block read is quicker.


 > If I read
 > characters using getc() one at a time, it is much slower than using
 > fread() which reads to a buffer... and then run down the buffer.


 >  NP> If you know where it is (I've forgotten how to find out) then
 >  NP> access to the data is easy.

 >   (grin) Use fread(buffer, 512, 1, filename)...

 I don't understand C, but I think you will find that the buffer yo are
 using in C is NOT the machine area that is set aside as per my
 statement.


 >  NP> IOW it reads a sector from the file regardless of whether your
 >  NP> program asks for 1 byte or a block.

 >   No, it doesn't.

 Why say no it doesn't and then say....

  >   What you say is probably true,

This C programing has twisted your brain completely :-)


 > but every time I call getc() it must
 > read the sector again, move the file pointer up one, and take the
   ^^^^^^^^^^^^^^^^^^^^^^

Exactly as i said.

 >  NP> IOW it reads a sector from the file regardless of whether your
 >  NP> program asks for 1 byte or a block.


Is it clear now :-)

Niels

* OLR 5.1 *  "Before God we are all equally wise Ä and equally foolish."
               Ä  Einstein

--- FMail/386 0.98