Re: master 0724c6dbdae: Use string-search instead of string-match[-p] when possible

Re: master 0724c6dbdae: Use string-search instead of string-match[-p] when possible

[Dmitry Gutov]

Hi Mathias,

On 26/03/2023 13:47, Mattias Engdegård wrote:
> diff --git a/lisp/progmodes/ruby-ts-mode.el b/lisp/progmodes/ruby-ts-mode.el
> index d077c43ba52..d1034d467ab 100644
> --- a/lisp/progmodes/ruby-ts-mode.el
> +++ b/lisp/progmodes/ruby-ts-mode.el
> @@ -469,7 +469,7 @@ non-nil."
>     (let* (first-call )
>       (while (and parent
>                   (setq first-call (treesit-node-parent parent))
> -                (string-match-p "call" (treesit-node-type first-call)))
> +                (string-search "call" (treesit-node-type first-call)))
>         (setq parent first-call))
>       (treesit-node-start (treesit-search-subtree parent "\\." nil t))))

Thanks for this, it uncovered an existing bug in the code.

Re: master 0724c6dbdae: Use string-search instead of string-match[-p] when possible

[Emanuel Berg]

Okay, I've never heard of this, because the byte-compiler
don't complain and it isn't in the docstrings either what
I can see.

I have 14 string-match[-p], so what is the rule when they
should be replaced by `string-search' and when not do that,
and why?

-- 
underground experts united
https://dataswamp.org/~incal

Re: master 0724c6dbdae: Use string-search instead of string-match[-p] when possible

[Yuan Fu]

> On Mar 26, 2023, at 4:35 PM, Emanuel Berg <incal@dataswamp.org> wrote:
> 
> Okay, I've never heard of this, because the byte-compiler
> don't complain and it isn't in the docstrings either what
> I can see.
> 
> I have 14 string-match[-p], so what is the rule when they
> should be replaced by `string-search' and when not do that,
> and why?

When the search string shouldn’t be interpreted as regexp, I presume?

Yuan

Re: master 0724c6dbdae: Use string-search instead of string-match[-p] when possible

[Mattias Engdegård]

27 mars 2023 kl. 01.05 skrev Dmitry Gutov <dgutov@yandex.ru>:

> Thanks for this, it uncovered an existing bug in the code.

Happy to hear that!

Re: master 0724c6dbdae: Use string-search instead of string-match[-p] when possible

[Robert Pluim]

>>>>> On Sun, 26 Mar 2023 06:47:54 -0400 (EDT), Mattias Engdegård <mattiase@acm.org> said:

    Mattias> branch: master
    Mattias> commit 0724c6dbdaef2c549409836ba4f7999e05aa31fe
    Mattias> Author: Mattias Engdegård <mattiase@acm.org>
    Mattias> Commit: Mattias Engdegård <mattiase@acm.org>

    Mattias>     Use string-search instead of string-match[-p] when possible
    
    Mattias>     * lisp/progmodes/c-ts-mode.el
    Mattias>     (c-ts-mode--standalone-parent-skip-preproc):
    Mattias>     * lisp/progmodes/ruby-ts-mode.el (ruby-ts--align-chain):
    Mattias>     * lisp/treesit.el (treesit-max-buffer-size)
    Mattias>     (treesit--check-manual-coverage):
    Mattias>     Use the much faster string-search when just searching for a substring.
    Mattias> ---
    Mattias>  lisp/progmodes/c-ts-mode.el    | 2 +-
    Mattias>  lisp/progmodes/ruby-ts-mode.el | 2 +-
    Mattias>  lisp/treesit.el                | 4 ++--
    Mattias>  3 files changed, 4 insertions(+), 4 deletions(-)

Hmm, we could do this using a compiler macro, no? Something like

(put 'string-match-p 'compiler-macro
     #'byte-optimize-string-match-p)

(defun byte-optimize-string-match-p (form r string &optional start)
  (if (and (stringp r)
           (string-equal r (regexp-quote r)))
      `(string-search ,r ,string ,start)
    form))

(I donʼt think we can do it unconditionally for `string-match', since
`string-search' doesnʼt set the match data.)

Robert
-- 

Re: master 0724c6dbdae: Use string-search instead of string-match[-p] when possible

[Mattias Engdegård]

29 mars 2023 kl. 17.23 skrev Robert Pluim <rpluim@gmail.com>:

> Hmm, we could do this using a compiler macro, no? Something like
> 
> (put 'string-match-p 'compiler-macro
>     #'byte-optimize-string-match-p)

No, that would be incorrect: string-match-p obeys case-fold-search, string-search doesn't.

Replacing string-match[-p] with string-search in the source also makes the code clearer.

Re: master 0724c6dbdae: Use string-search instead of string-match[-p] when possible

[Robert Pluim]

>>>>> On Wed, 29 Mar 2023 17:34:36 +0200, Mattias Engdegård <mattiase@acm.org> said:

    Mattias> 29 mars 2023 kl. 17.23 skrev Robert Pluim <rpluim@gmail.com>:
    >> Hmm, we could do this using a compiler macro, no? Something like
    >> 
    >> (put 'string-match-p 'compiler-macro
    >> #'byte-optimize-string-match-p)

    Mattias> No, that would be incorrect: string-match-p obeys case-fold-search, string-search doesn't.

Right, Iʼd forgotten about that. And a warning would be a bit
intrusive, I think.

    Mattias> Replacing string-match[-p] with string-search in the source also makes the code clearer.

Thatʼs definitely true.

Robert
--