Re: [PATCH 4/4] useless-if-before-free: skip non-matching lines early

[Jim Meyering]

On Tue, Jul 26, 2016 at 7:28 AM, Ján Tomko <address@hidden> wrote:
> Check if there is any if keyword on the currently
> processed line by a simple regex before matching
> against the more expensive capturing regex.
> ---
>  ChangeLog                        | 8 ++++++++
>  build-aux/useless-if-before-free | 3 +++
>  2 files changed, 11 insertions(+)
>
> diff --git a/ChangeLog b/ChangeLog
> index 64e2026..5aebb2f 100644
> --- a/ChangeLog
> +++ b/ChangeLog
> @@ -1,5 +1,13 @@
>  2016-07-26  Ján Tomko  <address@hidden>
>
> +       useless-if-before-free: skip non-matching lines early
> +
> +       Check if there is any if keyword on the currently
> +       processed line by a simple regex before matching
> +       against the more expensive capturing regex.
...
> +++ b/build-aux/useless-if-before-free
> @@ -129,6 +129,9 @@ sub is_NULL ($)
>            $err = EXIT_ERROR, next;
>        while (defined (my $line = <FH>))
>          {
> +          # Skip non-matching lines early to save time
> +          if (not $line =~ /\bif/) { next }

Thank you for the patch.
How much of an improvement did this give you?
I ran some quick tests (albeit using \bif\b) and an input file
created with this command run from gnulib source tree:

  cat lib/*.c > k; cat k k k k k k k > j; cat j j j j > k

That creates a file with about 127MB on 4 million lines,
and running this script on it prints 1736 lines.

Adding these lines (the syntax I prefer),

          $line =~ /\bif\b/
            or next;

changed the best-of-10 elapsed run time from 1.50s to 1.35s,
for a 10% (aka 0.15s) speed-up.