Re: [PATCH] sort: Speed up keyed field sorting significantly using memchr

[Collin Funk]

Chris Down <ch...@chrisdown.name> writes:

> diff --git a/src/sort.c b/src/sort.c
> index e20ed2c8f..aaab47b50 100644
> --- a/src/sort.c
> +++ b/src/sort.c
> @@ -1725,8 +1725,8 @@ begfield (struct line const *line, struct keyfield 
> const *key)
>    if (tab != TAB_DEFAULT)
>      while (ptr < lim && sword--)
>        {
> -        while (ptr < lim && *ptr != tab)
> -          ++ptr;
> +        char *sep = memchr (ptr, tab, (size_t) (lim - ptr));
> +        ptr = sep ? sep : lim;
>          if (ptr < lim)
>            ++ptr;
>        }
> @@ -1778,8 +1778,8 @@ limfield (struct line const *line, struct keyfield 
> const *key)
>    if (tab != TAB_DEFAULT)
>      while (ptr < lim && eword--)
>        {
> -        while (ptr < lim && *ptr != tab)
> -          ++ptr;
> +        char *sep = memchr (ptr, tab, (size_t) (lim - ptr));
> +        ptr = sep ? sep : lim;
>          if (ptr < lim && (eword || echar))
>            ++ptr;
>        }

Thanks for the patch and detailed benchmarks!

I just pushed a tiny patch to avoid the size_t casts. Nothing saying
your patch is wrong, it is just most of us prefer to avoid them.

Collin


>From 430822d5f71ff95bc813ddc254d559d808177d57 Mon Sep 17 00:00:00 2001
Message-ID: <430822d5f71ff95bc813ddc254d559d808177d57.1774320260.git.collin.fu...@gmail.com>
From: Collin Funk <collin.fu...@gmail.com>
Date: Mon, 23 Mar 2026 19:32:21 -0700
Subject: [PATCH] maint: remove some unnecessary casts

* src/sort.c (begfield, limfield): Remove size_t casts.
---
 src/sort.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/src/sort.c b/src/sort.c
index aaab47b50..510c674d9 100644
--- a/src/sort.c
+++ b/src/sort.c
@@ -1725,7 +1725,7 @@ begfield (struct line const *line, struct keyfield const *key)
   if (tab != TAB_DEFAULT)
     while (ptr < lim && sword--)
       {
-        char *sep = memchr (ptr, tab, (size_t) (lim - ptr));
+        char *sep = memchr (ptr, tab, lim - ptr);
         ptr = sep ? sep : lim;
         if (ptr < lim)
           ++ptr;
@@ -1778,7 +1778,7 @@ limfield (struct line const *line, struct keyfield const *key)
   if (tab != TAB_DEFAULT)
     while (ptr < lim && eword--)
       {
-        char *sep = memchr (ptr, tab, (size_t) (lim - ptr));
+        char *sep = memchr (ptr, tab, lim - ptr);
         ptr = sep ? sep : lim;
         if (ptr < lim && (eword || echar))
           ++ptr;
-- 
2.53.0