From: YASUOKA Masahiko <yasuoka@openbsd.org>
Subject: Re: develop/m4 fix make test
To: kn@openbsd.org
Cc: tb@openbsd.org, sthen@openbsd.org, ports@openbsd.org
Date: Thu, 22 May 2025 21:46:28 +0900

On Thu, 22 May 2025 19:58:48 +0900 (JST)
YASUOKA Masahiko <yasuoka@openbsd.org> wrote:
> On Thu, 22 May 2025 19:27:50 +0900 (JST)
> YASUOKA Masahiko <yasuoka@openbsd.org> wrote:
>> On Thu, 22 May 2025 09:30:54 +0000
>> Klemens Nanni <kn@openbsd.org> wrote:
>>>>>> + @result{}
>>>>>> + ifelse(sysval, [0], , [errprint([ skipping: shell does not send signal 9
>>>>>> + ])m4exit([77])])dnl
>>>>>> Index: patches/patch-tests_test-sys_wait_c
>>>>>> ===================================================================
>>>>>> RCS file: patches/patch-tests_test-sys_wait_c
>>>>>> diff -N patches/patch-tests_test-sys_wait_c
>>>>>> --- /dev/null	1 Jan 1970 00:00:00 -0000
>>>>>> +++ patches/patch-tests_test-sys_wait_c	22 May 2025 08:54:54 -0000
>>>>>> @@ -0,0 +1,13 @@
>>>>>> +workaround for #include puzzle
>>>>>
>>>>> Could you mention what fails and or how?
>>>>> Otherwise this is just a new puzzle for porters trying to understand the patch.
>> 
>> including <sys/time.h> is in an ad-hoc way.  I was not able to
>> understood the proper way to compose #include lines in m4 source
>> code.

I could find a proper fix.  The patch to tests/test-sys_wait.c is
replaced to patch to tests/tests/sys_select.in.h .


ok?

Index: Makefile
===================================================================
RCS file: /cvs/ports/devel/m4/Makefile,v
diff -u -p -r1.41 Makefile
--- Makefile	22 Apr 2025 18:18:15 -0000	1.41
+++ Makefile	22 May 2025 12:40:22 -0000
@@ -4,7 +4,7 @@ DISTNAME=	m4-1.4.19
 CATEGORIES=	devel
 SITES=		${SITE_GNU:=m4/}
 EXTRACT_SUFX=	.tar.xz
-REVISION=	0
+REVISION=	1
 
 HOMEPAGE=	https://www.gnu.org/software/m4/
 
@@ -17,6 +17,9 @@ CONFIGURE_STYLE= gnu
 CONFIGURE_ARGS=	--program-prefix="g" \
 		--disable-nls \
 		--without-libiconv-prefix
+
+# $HOME must exist for cvs that is used by tests/test-vc-list-files-cvs.sh
+PORTHOME=	${WRKDIR}
 
 post-install:
 	${INSTALL_DATA_DIR} ${PREFIX}/share/examples/gm4
Index: patches/patch-doc_m4_texi
===================================================================
RCS file: /cvs/ports/devel/m4/patches/patch-doc_m4_texi,v
diff -u -p -r1.4 patch-doc_m4_texi
--- patches/patch-doc_m4_texi	4 Dec 2024 09:51:29 -0000	1.4
+++ patches/patch-doc_m4_texi	22 May 2025 12:40:22 -0000
@@ -1,3 +1,7 @@
+Redirect stderr to /dev/null, since our /bin/sh prints "Killed" when a
+subprocess has died by a KILL signal.  Delete that message since that is
+out of the scope of the test.
+
 Index: doc/m4.texi
 --- doc/m4.texi.orig
 +++ doc/m4.texi
@@ -10,3 +14,12 @@ Index: doc/m4.texi
  @end direntry
  
  @titlepage
+@@ -6756,7 +6756,7 @@ ifdef(`__unix__', ,
+ ')m4exit(`77')')dnl
+ changequote(`[', `]')
+ @result{}
+-syscmd([/bin/sh -c 'kill -9 $$'; st=$?; test $st = 137 || test $st = 265])
++syscmd([/bin/sh -c 'kill -9 $$'2>/dev/null; st=$?; test $st = 137 || test $st = 265])
+ @result{}
+ ifelse(sysval, [0], , [errprint([ skipping: shell does not send signal 9
+ ])m4exit([77])])dnl
Index: patches/patch-tests_sys_select_in_h
===================================================================
RCS file: patches/patch-tests_sys_select_in_h
diff -N patches/patch-tests_sys_select_in_h
--- /dev/null	1 Jan 1970 00:00:00 -0000
+++ patches/patch-tests_sys_select_in_h	22 May 2025 12:40:22 -0000
@@ -0,0 +1,14 @@
+Our <sys/select.h> has definition of 'sigset_t'
+
+Index: tests/sys_select.in.h
+--- tests/sys_select.in.h.orig
++++ tests/sys_select.in.h
+@@ -111,7 +111,7 @@
+    undeclared error occurs in <pthread.h>.
+    Do this after the include_next (for the sake of OpenBSD 5.0) but before
+    the split double-inclusion guard (for the sake of Solaris).  */
+-#if !((defined __GLIBC__ || defined __CYGWIN__ || defined __KLIBC__) \
++#if !((defined __GLIBC__ || defined __CYGWIN__ || defined __KLIBC__ || defined __OpenBSD__) \
+       && !defined __UCLIBC__)
+ # include <signal.h>
+ #endif