more questions about efficient C code

[Chris Torek]

> This example
[ if ((fp = fopen(...)) == NULL) vs fp = fopen(...); if (fp == NULL) ]
> bothers me.  I'm not sure what all makes up the fopen subroutine,
> but the addition or subtraction of one measly assignment statement
> *has* to be negligible when compared to what goes on in fopen.
> Even if that code is in the innermost of inner loops, the "optimization"
> will still be unnoticable.  I know it's just an example of an
> assignment within a boolean expression, but I see a *lot* of programs
> with that same code.  Are we really gaining anything, or is it
> merely psychological?

Heck, 500 nanoseconds is significant, isn't it? :-)

Besides, you can save four bytes per test.  I'll bet it makes a disk
block or two of difference.  And we all need all the space we can get,
right? :-)

Seriously, the difference in the fopen call is pretty small, but
the difference in something like

	for (...)
		for (...) {
			if ((j = (k >> 2) - m) == 0)
				k++;
		}

can be important.  Once you get into the habit of testing the results
of an assignment, it doesn't even look funny anymore.
-- 
In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 4251)
UUCP:	seismo!umcp-cs!chris
CSNet:	chris at umcp-cs		ARPA:	chris at maryland