Comments on: To if-else-if or if-if, that is the question

By: The Shape of Code » Estimating variance when measuring source

The Shape of Code » Estimating variance when measuring source — Thu, 08 Oct 2009 19:25:26 +0000

[...] the percentage of equality tests that occur in the controlling expressions of if-if/if-else-if sequences and switch statements we get the [...]

By: The Shape of Code » Using Coccinelle to match if sequences

The Shape of Code » Using Coccinelle to match if sequences — Mon, 31 Aug 2009 16:06:06 +0000

[...] have been using Coccinelle to obtain measurements of various properties of C if and switch statements. It is rare to find a tool that does exactly what is desired but it is often [...]

By: Derek-Jones

Derek-Jones — Mon, 24 Aug 2009 15:16:34 +0000

More tests appear to be needed at runtime for the if-if usage. I say 'appear' because after gcc's impressive performance on 3n+1 I thought it might spot when if-if could be implemented as if-else-if; a quick check shows that with -O2 optimization gcc does indeed implement this transformation.

By: Safer Code

Safer Code — Mon, 24 Aug 2009 05:08:31 +0000

Hey Derek
I have a doubt when you say that there is no obvious cost or benefit…If the control variable is being changed then obviously we would have to go for the 2nd form because the first wouldn’t cover all cases and if we assume that the variable is not being changed then I agree there might not be any difference in worst case scenario (last condition holds true) but wouldn’t the 2nd form cause penalties in best case scenario (first condition holds true) as many more tests would be done in 2nd case even though we don’t need to check anymore..