[REBOL] Re: el(se)if constriction
From: joel:neely:fedex at: 28-Oct-2000 17:47
Hi, P-O (or is it PeO?)
Thanks for the numbers! It's really helpful to me to have this kind
of comparison of approaches to a task.
[rebol-bounce--rebol--com] wrote:
> Some results
>
> Data set: 70 files, with identical (but different) data to be parsed
> (tv program tables), about 20-50 entries in each of the files.
>
> The conditions are using parse to extract the start time, title &
> description, and title only (if no description was available).
>
> 3 tests 9 tests (6 extra of the most advanced)
> elif (/me) 0:51 1:13 - reduce is probably slowing it down
> pif (Ladislav) 0:50 1:11
> ifs (Andrew) 0:56 1:11
> any/if (Gabriele) 1:06 1:01 - No idea why 9 tests is faster than 3
>
> so, the any/if combination is both the slowest and the fastest one
> (and I did not mix up the test times between the 3 and 9 tests).
>
My habit is to rerun a benchmark test a few times (3-5) and average
the results. That tends to eliminate transient events that may
affect a specific timing, such as O/S flushing its buffers, REBOL
doing garbage collection, and the effect of other processes running
(if you're in an environment that supports multitasking).
-jn-
