From: Theo Buehler <tb@theobuehler.org>
Subject: Re: pgpool still broken; needs strchrnul() [Re: CVS: cvs.openbsd.org: ports]
To: Landry Breuil <landry@cvs.openbsd.org>, ports@openbsd.org
Date: Sat, 6 Jun 2026 14:51:22 +0200

On Sat, Jun 06, 2026 at 01:14:44PM +0100, Stuart Henderson wrote:
> On 2026/06/05 02:24, Landry Breuil wrote:
> > CVSROOT:	/cvs
> > Module name:	ports
> > Changes by:	landry@cvs.openbsd.org	2026/06/05 02:24:15
> > 
> > Modified files:
> > 	databases/pgpool: Makefile distinfo 
> > Removed files:
> > 	databases/pgpool/patches: patch-src_include_utils_fe_ports_h 
> > 	                          patch-src_include_watchdog_wd_utils_h 
> > 	                          patch-src_utils_json_writer_c 
> > 	                          patch-src_utils_ssl_utils_c 
> > 	                          patch-src_watchdog_wd_escalation_c 
> > 
> > Log message:
> > databases/pgpool: update to 4.6.7 and unbreak with llvm 22
> > 
> 
> this wants strchrnul() which we don't have - an implementation is
> available from Crystal though;
> https://research.exoticsilicon.com/downloads/strchrnul_patchset_7.8.sig)

There's a compat impl in src/parse/snprintf.c. I suggest we copy a
fixed version of that dance. Upstream should likely move all this into
the appropriate compat header.

(I discussed this with landry, not sure if he missed a cvs add)

The port could also use a CFLAGS += -Wno-unknown-warning-option...

diff --git a/databases/pgpool/patches/patch-src_utils_pool_process_reporting_c b/databases/pgpool/patches/patch-src_utils_pool_process_reporting_c
new file mode 100644
index 0000000000..a11f42dc8d
--- /dev/null
+++ b/databases/pgpool/patches/patch-src_utils_pool_process_reporting_c
@@ -0,0 +1,46 @@
+Index: src/utils/pool_process_reporting.c
+--- src/utils/pool_process_reporting.c.orig
++++ src/utils/pool_process_reporting.c
+@@ -45,6 +45,42 @@ static void write_one_field_v2(POOL_CONNECTION * front
+ static char *db_node_status(int node);
+ static char *db_node_role(int node);
+ 
++/*
++ * XXX - taken from src/parser/snprintf.c
++ * This really belongs into a header. That the snprintf version isn't
++ * const correct is also not the greatest of ideas.
++ */
++/*
++ * If strchrnul exists (it's a glibc-ism), it's a good bit faster than the
++ * equivalent manual loop.  If it doesn't exist, provide a replacement.
++ *
++ * Note: glibc declares this as returning "char *", but that would require
++ * casting away const internally, so we don't follow that detail.
++ */
++#ifndef HAVE_STRCHRNUL
++
++static inline char *
++strchrnul(const char *s, int c)
++{
++	while (*s != '\0' && *s != (char)c)
++		s++;
++	return (char *)s;
++}
++
++#else
++
++/*
++ * glibc's <string.h> declares strchrnul only if _GNU_SOURCE is defined.
++ * While we typically use that on glibc platforms, configure will set
++ * HAVE_STRCHRNUL whether it's used or not.  Fill in the missing declaration
++ * so that this file will compile cleanly with or without _GNU_SOURCE.
++ */
++#ifndef _GNU_SOURCE
++extern char *strchrnul(const char *s, int c);
++#endif
++
++#endif							/* HAVE_STRCHRNUL */
++
+ void
+ send_row_description(POOL_CONNECTION * frontend, POOL_CONNECTION_POOL * backend,
+ 					 short num_fields, char **field_names)