Re: [elpa] externals/compat 6a60af22da: Optimize string-search

[Philip Kaludercic]

ELPA Syncer <elpasync@gnu.org> writes:

> branch: externals/compat
> commit 6a60af22da1f1a7b08457a4016b9412a475afe4a
> Author: Daniel Mendler <mail@daniel-mendler.de>
> Commit: Daniel Mendler <mail@daniel-mendler.de>
>
>     Optimize string-search
> ---
>  compat-28.el | 5 ++---
>  1 file changed, 2 insertions(+), 3 deletions(-)
>
> diff --git a/compat-28.el b/compat-28.el
> index fe69315522..b305ab2fff 100644
> --- a/compat-28.el
> +++ b/compat-28.el
> @@ -52,9 +52,8 @@ issues are inherited."
>    (when (and start-pos (or (< (length haystack) start-pos)
>                             (< start-pos 0)))
>      (signal 'args-out-of-range (list start-pos)))
> -  (save-match-data
> -    (let ((case-fold-search nil))
> -      (string-match (regexp-quote needle) haystack start-pos))))
> +  (let (case-fold-search)
> +    (string-match-p (regexp-quote needle) haystack start-pos)))
>  
>  (compat-defun length= (sequence length) ;; [[compat-tests:length=]]
>    "Returns non-nil if SEQUENCE has a length equal to LENGTH."

I don't believe this is a legal optimisation, even if the regular
expression is quoted.  Consider

(progn
  (string-match (rx (* "a") (group (* "b")) (* "a")) "abba")
  (match-data))
; => (0 4 1 3)

and

(progn
  (string-match (rx (* "a") (group (* "b")) (* "a")) "abba")
  (string-search "foo" "foobar")
  (match-data))
; => (0 4 1 3)

but

(progn
  (string-match (rx (* "a") (group (* "b")) (* "a")) "abba")
  (let ((case-fold-search nil))
    (string-match (regexp-quote "foo") "foobar" 0))
  (match-data))
; => (0 3)