BL> I don't understand why fread() is faster than getc(). How do
 BL> they get the file data into the buffer? Don't you have to move
 BL> it one character at a time?

 RM> yes, but getc() does it with a call to a C function for *each*
 RM> character, instead of reading a bunch of chars with just one
 RM> function call.

  Actually, getc() is a macro that only calls the fgetc() function
when the fgetc() buffer is empty. I found this out for myself (after
asking Paul with no success) by looking in the stdio.h file.

  But I still don't know how fgetc() works (although Paul has told me
he has tested it a million times without discovering which is faster).
This information was of no value to me whatsoever (as usual).
 
  Paul *seems* to be saying that if I don't set the stream buffer with 
sefbuf() then fgetc() does not use a buffer. I've been trying to get 
Paul to be explicit, but as always when he's not sure he becomes 
contradictory. I can't make setbuf() work for me.

  I think what you have to do is use setvbuf() with the buffer named
NULL, which creates a buffer on the heap named the way fgetc() can use
it. But of course, Paul prefers to be obscure. I think he is that shit
who was always looking over your shoulder at school; the one who never
thought for himself and never knew anything of substance but knew all
the buzzwords... Rod Edwards, Paul Speed - his name is on the tip of
my bum. 

  Every test I run has getc() 3-5 times slower than the buffered
fread().  

 RM> I learned this the hard way with output - it's a lot faster to
 RM> write a whole string to the screen than it is to write the
 RM> string one char at a time.

  Yes... one of the first things I did in VB was to try various ways
of doing things in a loop.

Regards,
Bob
___ Blue Wave/QWK v2.12
@EOT:

---