Re: books/utp tiny MASTER_SITES correction
On Fri, Apr 14, 2017 at 08:25:46AM +0200, Andreas Kusalananda Kähäri wrote: > On Thu, Apr 13, 2017 at 07:23:41PM -0600, Anthony J. Bentley wrote: > > Stuart Henderson writes: > > > -HOMEPAGE = http://home.windstream.net/kollar/utp/ > > > - > > > > It's up for me... > > Weird, I get timeouts in Chrome and "Connection reset by peer" from > Curl, both from home and from work. But I can see that there's a recent > Google cache of the page (more recent than when I first started looking > for where the page had gone), so if it's actually up but difficult to > get to for just me, then I assume it's ok. > > > > > > -MASTER_SITES = http://ftp.ffii.org/pub/groff/contrib/documenta > > > tion/utp/ \ > > > - http://home.windstream.net/kollar/utp/ > > > +MASTER_SITES = https://groff.ffii.org/groff/contrib/documentat > > > ion/utp/ \ > > > + ftp://ftp.ffii.org/pub/groff/contrib/documentation/utp/ Also fails from Spain, Netherlands and Germany. I've reported the problem to their support email. -- Juan Francisco Cantero Hurtado http://juanfra.info
Re: news/nn overhaul
I don’t currently use OpenBSD on any of my main work machines, so I don’t use
the NN port right now, but I heartily approve of the overall simplification
enumerated below. Thank you!
--
Aaron W. Hsu | arcf...@sacrideo.us | http://www.sacrideo.us
Support my Open Work: http://www.gratipay.com/Co-dfns/
From: Christian Weisgerber
Sent: Friday, April 14, 2017 6:00 PM
To: ports@openbsd.org
Cc: Aaron W. Hsu
Subject: news/nn overhaul
Does anybody still use news/nn? Here's an overhaul of the port:
* Get rid of the MACHINE_ARCH dance and use a single m-openbsd.h
file. We only need to define a few basic types that are the same
on all our archs. NETWORK_BYTE_ORDER could be gotten from
, but it is only used with NETWORK_DATABASE, which we
don't define.
* Simplify the contorted substitutions in Makefile and use SUBST_CMD
where possible.
* clang throws a hissy fit over nested comments. Since fixing those
destroys the diff-ability of s-openbsd.h compared to s-template.h
anyway, I went ahead and compactified the file.
* clang on amd64 with our binutils ld produced a relocation error
for BC in term.c. I think these are globals that are already
defined in libtermlib, and thus should be declared extern.
Very lightly tested on arm64.
Index: Makefile
===
RCS file: /cvs/ports/news/nn/Makefile,v
retrieving revision 1.15
diff -u -p -r1.15 Makefile
--- Makefile1 Sep 2016 10:53:27 - 1.15
+++ Makefile14 Apr 2017 21:26:28 -
@@ -1,8 +1,8 @@
# $OpenBSD: Makefile,v 1.15 2016/09/01 10:53:27 jasper Exp $
-COMMENT = No News is Good News (Fast CLI Point & Read Newsreader)
+COMMENT = menu-based Usenet newsreader
DISTNAME = nn-6.7.3
-REVISION = 2
+REVISION = 3
CATEGORIES = news
HOMEPAGE = http://www.nndev.org
@@ -23,36 +23,12 @@ USE_GROFF = Yes
NO_TEST = Yes
MAKE_ENV = CC="${CC}"
-.if "${MACHINE_ARCH}" == "alpha"
-MFILE =m-alpha.h
-.elif "${MACHINE_ARCH}" == "hppa"
-MFILE =m-hp9000.h
-.elif "${MACHINE_ARCH}" == "m88k"
-MFILE =m-m88000.h
-.elif "${MACHINE_ARCH}" == "powerpc"
-MFILE =m-powerpc.h
-.elif "${MACHINE_ARCH}" == "mips64"
-MFILE =m-mips.h
-.elif "${MACHINE_ARCH}" == "sparc64"
-MFILE =m-sparc.h
-.else
-MFILE =m-i80386.h
-.endif
-
pre-configure:
- sed -e "s/OBSD_PREFIX/${PREFIX:S/\//\\\//g}/" \
- -e "s/m-i80386\.h/${MFILE}/" \
- ${FILESDIR}/config.h > ${WRKSRC}/config.h
- cp ${FILESDIR}/s-openbsd.h ${WRKSRC}/conf/s-openbsd.h
-
-pre-install:
- sed -e "s/\/usr\/local/${PREFIX:C/\//\\\//g}/" \
- ${WRKSRC}/inst > ${WRKSRC}/inst.new
- mv ${WRKSRC}/inst ${WRKSRC}/inst.old
- mv ${WRKSRC}/inst.new ${WRKSRC}/inst
- chmod 755 ${WRKSRC}/inst
+ ${SUBST_DATA} ${FILESDIR}/config.h ${WRKSRC}/config.h
+ cp ${FILESDIR}/[ms]-openbsd.h ${WRKSRC}/conf/
do-install:
- cd ${WRKSRC} && echo `pwd` && ./inst n
+ perl -pi.bak -e 's:${TRUEPREFIX}:${PREFIX}:g' ${WRKSRC}/inst
+ cd ${WRKSRC} && ./inst n
.include
Index: files/config.h
===
RCS file: /cvs/ports/news/nn/files/config.h,v
retrieving revision 1.1.1.1
diff -u -p -r1.1.1.1 config.h
--- files/config.h 17 Nov 2008 08:42:13 - 1.1.1.1
+++ files/config.h 14 Apr 2017 21:26:28 -
@@ -122,7 +122,7 @@
* conf/m-template.h as a starting point for writing you own.
*/
-#include "m-i80386.h"
+#include "m-openbsd.h"
/* OWNERSHIP ***
@@ -173,10 +173,10 @@
*(= LIB_DIRECTORY/Log if undefined).
*/
-#define BIN_DIRECTORY "OBSD_PREFIX/bin"
-#define LIB_DIRECTORY "OBSD_PREFIX/lib/nn"
-#define CLIENT_DIRECTORY "OBSD_PREFIX/libexec/nn"
-#define HELP_DIRECTORY "OBSD_PREFIX/share/doc/nn"
+#define BIN_DIRECTORY "${PREFIX}/bin"
+#define LIB_DIRECTORY "${PREFIX}/lib/nn"
+#define CLIENT_DIRECTORY "${PREFIX}/libexec/nn"
+#define HELP_DIRECTORY "${PREFIX}/share/doc/nn"
#define TMP_DIRECTORY "/tmp"
#define LOG_FILE "/var/log/nn"
@@ -240,13 +240,13 @@
* DAEMON_MAN - nnmaster
*/
-#define USER_MAN_DIR "OBSD_PREFIX/man/man1"
+#define USER_MAN_DIR "${PREFIX}/man/man1"
#define USER_MAN_SECTION "1"
-#define SYS_MAN_DIR"OBSD_PREFIX/man/man1"
+#define SYS_MAN_DIR"${PREFIX}/man/man1"
#define SYS_MAN_SECTION"1m"
-#define DAEMON_MAN_DIR "OBSD_PREFIX/man/man8"
+#define DAEMON_MAN_DIR "${PREFIX}/man/man8"
#define DAEMON_MAN_SECTION "8"
Index: files/m-openbsd.h
===
RCS file: files/m-openbsd.h
diff -N files/m-openbsd.h
--- /dev/null
Fix for ruby 2.4 heap overflow
This fixes the ruby 2.4 heap overflow reported by
timo.my...@wickedbsd.net. The underlying cause was they renamed
variables in gc.c but forgot to update the configure script.
Ruby bug report: https://bugs.ruby-lang.org/issues/13438
Tested briefly on amd64, I'll run more tests soon. Assuming no
problems I'll be committing this weekend unless I hear objections.
Thanks,
Jeremy
Index: Makefile
===
RCS file: /cvs/ports/lang/ruby/2.4/Makefile,v
retrieving revision 1.3
diff -u -p -r1.3 Makefile
--- Makefile24 Mar 2017 14:39:05 - 1.3
+++ Makefile14 Apr 2017 23:16:12 -
@@ -16,6 +16,8 @@ PKGNAME-ri_docs = ruby24-ri_docs-${VERSI
PKG_ARCH-ri_docs = *
WANTLIB-ri_docs = # empty
+REVISION-main =0
+
NEXTVER = 2.5
PKGSPEC-main = ruby->=${RUBYLIBREV},<${NEXTVER}
Index: patches/patch-configure
===
RCS file: /cvs/ports/lang/ruby/2.4/patches/patch-configure,v
retrieving revision 1.2
diff -u -p -r1.2 patch-configure
--- patches/patch-configure 24 Mar 2017 14:39:05 - 1.2
+++ patches/patch-configure 14 Apr 2017 23:15:58 -
@@ -5,8 +5,10 @@ Fix so name, checking for DOT and DOXYGE
Override the arch setting to remove OpenBSD version from it,
so ports don't have to be bumped when OpenBSD version changes.
+Rename HEAP_ALIGN_LOG to HEAP_PAGE_ALIGN_LOG to fix heap overflow.
+
--- configure.orig Tue Mar 21 22:56:10 2017
-+++ configure Wed Mar 22 08:33:15 2017
configure Fri Apr 14 16:15:30 2017
@@ -424,7 +424,7 @@ if test -z "$2"; then :
else
@@ -16,6 +18,21 @@ so ports don't have to be bumped when Op
if test "$FOLD"; then :
echo "$2" | fold -s -w50 |
+@@ -19464,12 +19464,12 @@ fi
+ colorize_result "$rb_cv_page_size_log" ; }
+ if test $rb_cv_page_size_log != no; then
+ cat >>confdefs.h <<_ACEOF
+-#define HEAP_ALIGN_LOG $rb_cv_page_size_log
++#define HEAP_PAGE_ALIGN_LOG $rb_cv_page_size_log
+ _ACEOF
+
+ else
+ cat >>confdefs.h <<_ACEOF
+-#define HEAP_ALIGN_LOG 12
++#define HEAP_PAGE_ALIGN_LOG 12
+ _ACEOF
+
+ fi
@@ -25601,7 +25601,7 @@ esac
openbsd*|mirbsd*) :
[PATCH] curses support for qemu-2.8.1
Hello,
QEMU's -curses option is not enabled because configure script requires
the information of ncurses by pkg-config.
The diff specifies the place of ncurses library to enable QEMU's
curses support.
ok or comment?
--
SASANO Takayoshi (JG1UAA)
Index: Makefile
===
RCS file: /cvs/ports/emulators/qemu/Makefile,v
retrieving revision 1.159
diff -u -p -r1.159 Makefile
--- Makefile10 Apr 2017 13:13:02 - 1.159
+++ Makefile14 Apr 2017 13:49:20 -
@@ -78,6 +78,7 @@ CONFIGURE_ARGS= --prefix=${PREFIX} \
--host-cc="${CC}" \
--extra-cflags="${EXTRA_CFLAGS}" \
--extra-ldflags="${EXTRA_LDFLAGS}" \
+ --enable-curses \
--disable-bsd-user \
--disable-gcrypt \
--disable-smartcard \
Index: patches/patch-configure
===
RCS file: /cvs/ports/emulators/qemu/patches/patch-configure,v
retrieving revision 1.50
diff -u -p -r1.50 patch-configure
--- patches/patch-configure 22 Dec 2016 09:05:06 - 1.50
+++ patches/patch-configure 14 Apr 2017 13:49:20 -
@@ -1,7 +1,64 @@
$OpenBSD: patch-configure,v 1.50 2016/12/22 09:05:06 ajacoutot Exp $
configure.orig Tue Dec 6 22:54:54 2016
-+++ configure Sat Dec 10 23:54:52 2016
-@@ -4740,10 +4740,6 @@ fi
+--- configure.orig Fri Mar 31 23:07:03 2017
configure Fri Apr 14 06:19:09 2017
+@@ -2924,54 +2924,8 @@ fi
+ ##
+ # curses probe
+ if test "$curses" != "no" ; then
+- if test "$mingw32" = "yes" ; then
+-curses_inc_list="$($pkg_config --cflags ncurses 2>/dev/null):"
+-curses_lib_list="$($pkg_config --libs ncurses 2>/dev/null):-lpdcurses"
+- else
+-curses_inc_list="$($pkg_config --cflags ncursesw
2>/dev/null):-I/usr/include/ncursesw:"
+-curses_lib_list="$($pkg_config --libs ncursesw
2>/dev/null):-lncursesw:-lcursesw"
+- fi
+- curses_found=no
+- cat > $TMPC << EOF
+-#include
+-#include
+-#include
+-int main(void) {
+- const char *s = curses_version();
+- wchar_t wch = L'w';
+- setlocale(LC_ALL, "");
+- resize_term(0, 0);
+- addwstr(L"wide chars\n");
+- addnwstr(&wch, 1);
+- add_wch(WACS_DEGREE);
+- return s != 0;
+-}
+-EOF
+- IFS=:
+- for curses_inc in $curses_inc_list; do
+-IFS=:
+-for curses_lib in $curses_lib_list; do
+- unset IFS
+- if compile_prog "$curses_inc" "$curses_lib" ; then
+-curses_found=yes
+-QEMU_CFLAGS="$curses_inc $QEMU_CFLAGS"
+-libs_softmmu="$curses_lib $libs_softmmu"
+-break
+- fi
+-done
+-if test "$curses_found" = yes ; then
+- break
+-fi
+- done
+- unset IFS
+- if test "$curses_found" = "yes" ; then
+-curses=yes
+- else
+-if test "$curses" = "yes" ; then
+- feature_not_found "curses" "Install ncurses devel"
+-fi
+-curses=no
+- fi
++ curses=yes
++ libs_softmmu="-lncurses $libs_softmmu"
+ fi
+
+ ##
+@@ -4740,10 +4694,6 @@ fi
if test "$gcov" = "yes" ; then
CFLAGS="-fprofile-arcs -ftest-coverage -g $CFLAGS"
LDFLAGS="-fprofile-arcs -ftest-coverage $LDFLAGS"
@@ -12,7 +69,7 @@ $OpenBSD: patch-configure,v 1.50 2016/12
fi
##
-@@ -4892,7 +4888,7 @@ if test \( "$cpu" = "i386" -o "$cpu" = "x86_64" \) -a
+@@ -4892,7 +4842,7 @@ if test \( "$cpu" = "i386" -o "$cpu" = "x86_64" \) -a
# Different host OS linkers have different ideas about the name of the ELF
# emulation. Linux and OpenBSD use 'elf_i386'; FreeBSD uses the _fbsd
# variant; and Windows uses i386pe.
news/nn overhaul
Does anybody still use news/nn? Here's an overhaul of the port:
* Get rid of the MACHINE_ARCH dance and use a single m-openbsd.h
file. We only need to define a few basic types that are the same
on all our archs. NETWORK_BYTE_ORDER could be gotten from
, but it is only used with NETWORK_DATABASE, which we
don't define.
* Simplify the contorted substitutions in Makefile and use SUBST_CMD
where possible.
* clang throws a hissy fit over nested comments. Since fixing those
destroys the diff-ability of s-openbsd.h compared to s-template.h
anyway, I went ahead and compactified the file.
* clang on amd64 with our binutils ld produced a relocation error
for BC in term.c. I think these are globals that are already
defined in libtermlib, and thus should be declared extern.
Very lightly tested on arm64.
Index: Makefile
===
RCS file: /cvs/ports/news/nn/Makefile,v
retrieving revision 1.15
diff -u -p -r1.15 Makefile
--- Makefile1 Sep 2016 10:53:27 - 1.15
+++ Makefile14 Apr 2017 21:26:28 -
@@ -1,8 +1,8 @@
# $OpenBSD: Makefile,v 1.15 2016/09/01 10:53:27 jasper Exp $
-COMMENT = No News is Good News (Fast CLI Point & Read Newsreader)
+COMMENT = menu-based Usenet newsreader
DISTNAME = nn-6.7.3
-REVISION = 2
+REVISION = 3
CATEGORIES = news
HOMEPAGE = http://www.nndev.org
@@ -23,36 +23,12 @@ USE_GROFF = Yes
NO_TEST = Yes
MAKE_ENV = CC="${CC}"
-.if "${MACHINE_ARCH}" == "alpha"
-MFILE =m-alpha.h
-.elif "${MACHINE_ARCH}" == "hppa"
-MFILE =m-hp9000.h
-.elif "${MACHINE_ARCH}" == "m88k"
-MFILE =m-m88000.h
-.elif "${MACHINE_ARCH}" == "powerpc"
-MFILE =m-powerpc.h
-.elif "${MACHINE_ARCH}" == "mips64"
-MFILE =m-mips.h
-.elif "${MACHINE_ARCH}" == "sparc64"
-MFILE =m-sparc.h
-.else
-MFILE =m-i80386.h
-.endif
-
pre-configure:
- sed -e "s/OBSD_PREFIX/${PREFIX:S/\//\\\//g}/" \
- -e "s/m-i80386\.h/${MFILE}/" \
- ${FILESDIR}/config.h > ${WRKSRC}/config.h
- cp ${FILESDIR}/s-openbsd.h ${WRKSRC}/conf/s-openbsd.h
-
-pre-install:
- sed -e "s/\/usr\/local/${PREFIX:C/\//\\\//g}/" \
- ${WRKSRC}/inst > ${WRKSRC}/inst.new
- mv ${WRKSRC}/inst ${WRKSRC}/inst.old
- mv ${WRKSRC}/inst.new ${WRKSRC}/inst
- chmod 755 ${WRKSRC}/inst
+ ${SUBST_DATA} ${FILESDIR}/config.h ${WRKSRC}/config.h
+ cp ${FILESDIR}/[ms]-openbsd.h ${WRKSRC}/conf/
do-install:
- cd ${WRKSRC} && echo `pwd` && ./inst n
+ perl -pi.bak -e 's:${TRUEPREFIX}:${PREFIX}:g' ${WRKSRC}/inst
+ cd ${WRKSRC} && ./inst n
.include
Index: files/config.h
===
RCS file: /cvs/ports/news/nn/files/config.h,v
retrieving revision 1.1.1.1
diff -u -p -r1.1.1.1 config.h
--- files/config.h 17 Nov 2008 08:42:13 - 1.1.1.1
+++ files/config.h 14 Apr 2017 21:26:28 -
@@ -122,7 +122,7 @@
* conf/m-template.h as a starting point for writing you own.
*/
-#include "m-i80386.h"
+#include "m-openbsd.h"
/* OWNERSHIP ***
@@ -173,10 +173,10 @@
*(= LIB_DIRECTORY/Log if undefined).
*/
-#define BIN_DIRECTORY "OBSD_PREFIX/bin"
-#define LIB_DIRECTORY "OBSD_PREFIX/lib/nn"
-#define CLIENT_DIRECTORY "OBSD_PREFIX/libexec/nn"
-#define HELP_DIRECTORY "OBSD_PREFIX/share/doc/nn"
+#define BIN_DIRECTORY "${PREFIX}/bin"
+#define LIB_DIRECTORY "${PREFIX}/lib/nn"
+#define CLIENT_DIRECTORY "${PREFIX}/libexec/nn"
+#define HELP_DIRECTORY "${PREFIX}/share/doc/nn"
#define TMP_DIRECTORY "/tmp"
#define LOG_FILE "/var/log/nn"
@@ -240,13 +240,13 @@
* DAEMON_MAN - nnmaster
*/
-#define USER_MAN_DIR "OBSD_PREFIX/man/man1"
+#define USER_MAN_DIR "${PREFIX}/man/man1"
#define USER_MAN_SECTION "1"
-#define SYS_MAN_DIR"OBSD_PREFIX/man/man1"
+#define SYS_MAN_DIR"${PREFIX}/man/man1"
#define SYS_MAN_SECTION"1m"
-#define DAEMON_MAN_DIR "OBSD_PREFIX/man/man8"
+#define DAEMON_MAN_DIR "${PREFIX}/man/man8"
#define DAEMON_MAN_SECTION "8"
Index: files/m-openbsd.h
===
RCS file: files/m-openbsd.h
diff -N files/m-openbsd.h
--- /dev/null 1 Jan 1970 00:00:00 -
+++ files/m-openbsd.h 14 Apr 2017 21:26:28 -
@@ -0,0 +1,8 @@
+/* $OpenBSD */
+
+/* MACHINE TYPE DEFINED TYPEVALUE RANGE */
+
+typedef unsigned char int8; /*0 .. 255 */
+typedef short int16; /* -10,000 .. 10,000 */
+typedef intint32; /* -100,000 .. 100,000 */
+typedef unsigned int uint32; /*0 ..
devel/boehm-gc: add aarch64 support
devel/boehm-gc compiles on arm64, but it doesn't actually work.
The regression tests all fail.
There is support for aarch64, but not on OpenBSD. The patch below
copies the required pieces into place. With this, the regression
tests all succeed.
ok?
Index: Makefile
===
RCS file: /cvs/ports/devel/boehm-gc/Makefile,v
retrieving revision 1.64
diff -u -p -r1.64 Makefile
--- Makefile10 Apr 2017 11:45:25 - 1.64
+++ Makefile14 Apr 2017 16:26:07 -
@@ -10,6 +10,7 @@ DISTNAME= gc-${VERSION}
PKGNAME-atomic=libatomic_ops-${LIBAO_VERSION}
PKGNAME-main= boehm-gc-${VERSION}
+REVISION-main= 0
DISTFILES= ${DISTNAME}${EXTRACT_SUFX} \
libatomic_ops-${LIBAO_VERSION}${EXTRACT_SUFX}
Index: patches/patch-include_private_gcconfig_h
===
RCS file: /cvs/ports/devel/boehm-gc/patches/patch-include_private_gcconfig_h,v
retrieving revision 1.21
diff -u -p -r1.21 patch-include_private_gcconfig_h
--- patches/patch-include_private_gcconfig_h12 Nov 2016 10:33:04 -
1.21
+++ patches/patch-include_private_gcconfig_h14 Apr 2017 16:26:07 -
@@ -2,10 +2,32 @@ $OpenBSD: patch-include_private_gcconfig
- All OpenBSD mips platforms are 64-bit, so correct the wordsize and alignment
- use __data_start instead of _fdata on OpenBSD/mips64
+- add support for OpenBSD/aarch64
--- include/private/gcconfig.h.origTue Aug 2 21:36:14 2016
-+++ include/private/gcconfig.h Thu Nov 10 04:14:50 2016
-@@ -1712,14 +1712,15 @@
include/private/gcconfig.h Fri Apr 14 17:44:40 2017
+@@ -102,7 +102,8 @@
+ # endif
+ # if defined(__aarch64__)
+ #define AARCH64
+-#if !defined(LINUX) && !defined(DARWIN) && !defined(FREEBSD)
++#if !defined(LINUX) && !defined(DARWIN) && !defined(FREEBSD) \
++ && !defined(OPENBSD)
+ # define NOSYS
+ # define mach_type_known
+ #endif
+@@ -137,6 +138,10 @@
+ #define ARM32
+ #define mach_type_known
+ # endif
++# if defined(OPENBSD) && defined(__aarch64__)
++#define AARCH64
++#define mach_type_known
++# endif
+ # if defined(OPENBSD) && defined(__sh__)
+ #define SH
+ #define mach_type_known
+@@ -1712,14 +1717,15 @@
# endif
# ifdef OPENBSD
#define OS_TYPE "OPENBSD"
@@ -24,3 +46,23 @@ $OpenBSD: patch-include_private_gcconfig
extern int _end[];
#define DATAEND ((ptr_t)(&_end))
#define DYNAMIC_LOADING
+@@ -2165,6 +2171,19 @@
+ extern char etext[];
+ # define DATASTART GC_FreeBSDGetDataStart(0x1000, (ptr_t)etext)
+ # define DATASTART_USES_BSDGETDATASTART
++# endif
++# ifdef OPENBSD
++# define OS_TYPE "OPENBSD"
++# ifndef GC_OPENBSD_THREADS
++# include
++# include
++# define STACKBOTTOM ((ptr_t)USRSTACK)
++# endif
++ extern int __data_start[];
++# define DATASTART ((ptr_t)__data_start)
++ extern int _end[];
++# define DATAEND ((ptr_t)(&_end))
++# define DYNAMIC_LOADING
+ # endif
+ # ifdef NOSYS
+ /* __data_start is usually defined in the target linker script. */
--
Christian "naddy" Weisgerber na...@mips.inka.de
Re: amd64-clang port build failures
Sebastien Marie:
> I am also surprised that lang/rust succeed (and devel/cargo didn't).
lang/rust builds with gcc 4.9, so it sidesteps clang.
> Is it possible to have more information on the build configuration ?
> Stuart already mentioned libestdc++ vs libc++ for example.
I have the base clang installed in addition to gcc:
/usr/bin/clang, /usr/bin/clang++
(I'm reluctant to publicly post instructions how to set this up,
since I know people who don't understand what they are doing will
mess with it and shoot themselves in the foot.)
Beyond that, I have the following ports infrastructure patches:
* Link clang,clang++ as cc,c++ into the leading PATH directory.
* Move amd64 from GCC4_ARCHS to CLANG_ARCHS.
* In the clang module, also link /usr/local/bin/clang and clang++
into the leading PATH directory.
Ports that use MODCLANG still use the ports clang with this.
If I didn't miss something, base clang++ uses libc++ and ports
clang++ still uses libestdc++.
Index: lang/clang/clang.port.mk
===
RCS file: /cvs/ports/lang/clang/clang.port.mk,v
retrieving revision 1.21
diff -u -p -r1.21 clang.port.mk
--- lang/clang/clang.port.mk14 Mar 2017 17:30:30 - 1.21
+++ lang/clang/clang.port.mk12 Apr 2017 22:13:01 -
@@ -31,10 +31,10 @@ _MODCLANG_LINKS =
.if ${_MODCLANG_ARCH_USES:L} == "yes"
BUILD_DEPENDS += devel/llvm>=${MODCLANG_VERSION}
-_MODCLANG_LINKS = clang gcc clang cc
+_MODCLANG_LINKS = clang gcc clang cc clang clang
. if ${MODCLANG_LANGS:L:Mc++}
-_MODCLANG_LINKS += clang++ g++ clang++ c++
+_MODCLANG_LINKS += clang++ g++ clang++ c++ clang++ clang++
# uses libestdc++
MODULES += gcc4
MODCLANG_CPPLIBDEP = ${MODGCC4_CPPLIBDEP}
Index: infrastructure/mk/arch-defines.mk
===
RCS file: /cvs/ports/infrastructure/mk/arch-defines.mk,v
retrieving revision 1.33
diff -u -p -r1.33 arch-defines.mk
--- infrastructure/mk/arch-defines.mk 10 Apr 2017 11:18:28 - 1.33
+++ infrastructure/mk/arch-defines.mk 12 Apr 2017 23:06:01 -
@@ -19,8 +19,8 @@ APM_ARCHS = amd64 i386 loongson macppc s
BE_ARCHS = hppa m88k mips64 powerpc sparc64
LE_ARCHS = aarch64 alpha amd64 arm i386 mips64el sh
LP64_ARCHS = aarch64 alpha amd64 sparc64 mips64 mips64el
-CLANG_ARCHS = aarch64
-GCC4_ARCHS = alpha amd64 arm armv7 i386 hppa landisk loongson \
+CLANG_ARCHS = aarch64 amd64
+GCC4_ARCHS = alpha arm armv7 i386 hppa landisk loongson \
macppc mips64 mips64el octeon powerpc sgi sh socppc sparc64
GCC3_ARCHS = aviion luna88k m88k
# XXX easier for ports that depend on mono
Index: infrastructure/mk/bsd.port.mk
===
RCS file: /cvs/ports/infrastructure/mk/bsd.port.mk,v
retrieving revision 1.1340
diff -u -p -r1.1340 bsd.port.mk
--- infrastructure/mk/bsd.port.mk 11 Apr 2017 15:36:56 - 1.1340
+++ infrastructure/mk/bsd.port.mk 12 Apr 2017 22:10:32 -
@@ -2437,6 +2437,11 @@ ${_WRKDIR_COOKIE}:
@ln -sf msgfmt ${WRKDIR}/bin/${name}
. endfor
.endif
+# force clang as default compiler
+ @ln -sf /usr/bin/clang ${WRKDIR}/bin/cc
+ @ln -sf /usr/bin/clang ${WRKDIR}/bin/gcc
+ @ln -sf /usr/bin/clang++ ${WRKDIR}/bin/c++
+ @ln -sf /usr/bin/clang++ ${WRKDIR}/bin/g++
@${_MAKE_COOKIE} $@
${_EXTRACT_COOKIE}: ${_WRKDIR_COOKIE}
--
Christian "naddy" Weisgerber na...@mips.inka.de
Re: [NEW] emulators/ucon64
On 2017-04-13 15:55:06, Frederic Cambus wrote: > Hi ports@, > > Here is a new port: emulators/ucon64 > > Comments? OK? > > >From DESCR: > > uCON64 is the emulator Swiss Army knife program. > > It supports almost every video game system (Consoles, Handheld, Arcade), > as well as all common patch file formats like IPS (with RLE compression), > APS, BSL (Baseline Patch format), PPF (Playstation Patch File), and Game > Genie. > Compiles and runs fine for me on amd64 -CURRENT. -- Bryan
[NEW] graphics/libsixel
Hi ports@, Here is a new port: graphics/libsixel Comments? OK? >From DESCR: This package provides encoder/decoder implementation for DEC SIXEL graphics, and some converter programs. SIXEL is one of image formats for printer and terminal imaging introduced by Digital Equipment Corp. (DEC). Its data scheme is represented as a terminal- friendly escape sequence. libsixel.tar.gz Description: application/tar-gz
transcode: fix RUN_DEPENDS / toolame: remove?
transcode wants to run the "twolame" executable from audio/twolame
instead of "toolame" provided by audio/toolame port.
ritchie /usr/ports/pobj/transcode-1.1.7/transcode-1.1.7$ ag twolame
export/export_toolame.c
27:#define MOD_NAME"export_twolame.so"
42:#define MOD_PRE twolame
88:/* check for twolame */
89:if (tc_test_program("twolame") != 0) return (TC_EXPORT_ERROR);
126:/* twolame command line */
129: "twolame -s %0.3f -b %d -m %c - \"%s\" 2>/dev/null %s",
This makes sense because twolame is an evolution of toolame which
doesn't appear maintained any more. The cli interface *seems* to be
backward compatible. Without twolame installed, transcode exits with
this error message when trying to use twolame:
[transcode] warning: The 'twolame' program could not be found.
I did not find an appropriate transcode magic to successfully test mp2
output, but the diff below fixes RUN_DEPENDS anyway. ok?
transcode was the last user of audio/toolame which doesn't seem to have
an upstream any more, so I propose to delete toolame. ok?
Index: Makefile
===
RCS file: /d/cvs/ports/multimedia/transcode/Makefile,v
retrieving revision 1.63
diff -u -p -r1.63 Makefile
--- Makefile10 Apr 2017 11:46:23 - 1.63
+++ Makefile14 Apr 2017 14:04:35 -
@@ -3,7 +3,7 @@
COMMENT= video stream processing tools
DISTNAME= transcode-1.1.7
-REVISION= 7
+REVISION= 8
CATEGORIES=multimedia
MASTER_SITES= http://cdn.bitbucket.org/france/transcode-tcforge/downloads/
EXTRACT_SUFX= .tar.bz2
@@ -56,7 +56,7 @@ LIB_DEPENDS= audio/lame \
RUN_DEPENDS= audio/sox \
audio/vorbis-tools \
- audio/toolame \
+ audio/twolame \
graphics/ffmpeg \
x11/mplayer
--
jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE
ruby-2.4 chunk canary corrupted
Hi,
Just getting back to ruby coding and noticed it doesn't seem to clean up
correctly when malloc flag 'C' is enabled.
When ruby process is aborted, like starting "ruby" from cli and pressing ^D it
prints out:
tmy@phobos home $ ruby
ruby(90150) in free(): chunk canary corrupted 0x11c714984000 0x3fd8@0x3fd8
Abort trap
tmy@phobos home $
Here's the backtrace for it.
(gdb) bt
#0 0x10efdd17157a in thrkill () at {standard input}:5
#1 0x10efdd1a7db9 in *_libc_abort () at /usr/src/lib/libc/stdlib/abort.c:52
#2 0x10efdd1a9bf6 in wrterror (d=0x7f7ef2e0, msg=0x10efdd2e7718 "chunk
canary corrupted %p %#tx@%#zx") at /usr/src/lib/libc/stdlib/malloc.c:306
#3 0x10efdd1a9cd2 in validate_canary (d=Variable "d" is not available.
) at /usr/src/lib/libc/stdlib/malloc.c:1053
#4 0x10efdd1ab289 in ofree (argpool=0x10ef5ef7dab0, p=0x10ef18984000,
clear=0, check=0, argsz=0) at /usr/src/lib/libc/stdlib/malloc.c:1359
#5 0x10efdd1ab563 in free (ptr=0x10ef18984000) at
/usr/src/lib/libc/stdlib/malloc.c:1439
#6 0x10ef8c73d038 in heap_page_free () from /usr/local/lib/libruby24.so.1.0
#7 0x10ef8c746ece in rb_objspace_free () from
/usr/local/lib/libruby24.so.1.0
#8 0x10ef8c83d6c0 in ruby_vm_destruct () from
/usr/local/lib/libruby24.so.1.0
#9 0x10ef8c7313e6 in ruby_cleanup () from /usr/local/lib/libruby24.so.1.0
#10 0x10ef8c731562 in ruby_run_node () from /usr/local/lib/libruby24.so.1.0
#11 0x10ed0e200670 in main () from /usr/local/bin/ruby24
Current language: auto; currently asm
Timo
Re: amd64-clang port build failures
On Fri, Apr 14, 2017 at 12:28:10AM +0100, Stuart Henderson wrote: > On 2017/04/13 22:33, Christian Weisgerber wrote: > > As you may know, on the arm64 architecture we use clang 4.0 as the > > system compiler. Other architectures for which clang support is > > available, like amd64, will follow sooner or later. > > > > Currently, a lot of the ports tree fails to build with clang. > > This must be fixed. > > > > To get us going, I have started running test builds on amd64 with > > clang as the compiler. Logs of the build failures are available > > here: > > http://build-failures.rhaalovely.net/amd64-clang/ > > > > Below is also a list of the 324 ports that broke in the first run. > > See if you recognize something and want to fix it... > > Thanks, that's a bit quicker than the SoftIron :) Interesting, boost > succeeded here. Was this using libestdc++ or libc++? > I am also surprised that lang/rust succeed (and devel/cargo didn't). Is it possible to have more information on the build configuration ? Stuart already mentioned libestdc++ vs libc++ for example. It would be interesting to be able to reproduce the build environment for testing. Some parts could be tricky and depend on proper compiler detection. Thanks. -- Sebastien Marie
Re: amd64-clang port build failures
Stuart Henderson:
> Interesting, boost succeeded here. Was this using libestdc++ or
> libc++?
No, boost was skipped completely due to ONLY_FOR_ARCHS=${GCC4_ARCHS}.
--
Christian "naddy" Weisgerber na...@mips.inka.de
Re: books/utp tiny MASTER_SITES correction
On 2017/04/14 08:25, Andreas Kusalananda Kähäri wrote: > On Thu, Apr 13, 2017 at 07:23:41PM -0600, Anthony J. Bentley wrote: > > Stuart Henderson writes: > > > -HOMEPAGE = http://home.windstream.net/kollar/utp/ > > > - > > > > It's up for me... > > Weird, I get timeouts in Chrome and "Connection reset by peer" from > Curl, both from home and from work. But I can see that there's a recent > Google cache of the page (more recent than when I first started looking > for where the page had gone), so if it's actually up but difficult to > get to for just me, then I assume it's ok. Traceroute gets close but no ping or http. Perhaps they've built a wall to block those dodgy un-american IP addresses that we're coming from.
Re: UPDATE: graphics/slop
Iirc our cmake creates files with .so and no version if you don't specify
SHARED_LIBS. Try setting that for the shared libraries name (version 0.0)
and rebuild/regen plist.
--
Sent from a phone, apologies for poor formatting.
On 14 April 2017 02:30:23 Jeremie Courreges-Anglas wrote:
Rafael Sadowski writes:
On Thu Apr 13, 2017 at 06:47:56PM +0100, Stuart Henderson wrote:
On 2017/04/13 19:19, Landry Breuil wrote:
> As for using gcc4 module for c++11 support, i'm a bit lost in the clang
> vs gcc4 vs atomics vs c++11 support way of dealing with things...
If it's for c++11 support you could use
MODULES=gcc4
MODGCC4_ARCHS= ${GCC3_ARCHS} ${GCC4_ARCHS}
MODGCC4_LANGS= c++
I've been trying to think of a nice abstraction so we don't need to
copy that boilerplate all over the place. Perhaps a USE_CXX11 flag?
+1 ... would be great for many ports! Until then:
Index: Makefile
===
RCS file: /cvs/ports/graphics/slop/Makefile,v
retrieving revision 1.10
diff -u -p -u -p -r1.10 Makefile
--- Makefile10 Apr 2017 11:46:21 - 1.10
+++ Makefile13 Apr 2017 18:12:20 -
@@ -1,21 +1,31 @@
# $OpenBSD: Makefile,v 1.10 2017/04/10 11:46:21 sthen Exp $
-V =4.1.16
+V =5.3.37
COMMENT = query for a selection and print to stdout (select operation)
DISTNAME = slop-${V}
CATEGORIES = graphics x11
-REVISION = 0
GH_ACCOUNT = naelstrof
GH_PROJECT = slop
GH_TAGNAME = v${V}
+MAINTAINER = Rafael Sadowski
+
# GPLv3+
PERMIT_PACKAGE_CDROM = Yes
-WANTLIB += ICE SM X11 Xext c m ${LIBCXX}
+WANTLIB += GL GLU ICE SM X11 Xext Xrender c m pthread ${LIBCXX}
+
+MODULES = devel/cmake \
+ gcc4
+
+MODGCC4_LANGS =c++
+MODGCC4_ARCHS =${GCC3_ARCHS} ${GCC4_ARCHS}
+
+BUILD_DEPENDS += graphics/glm
-MODULES = devel/cmake
+CONFIGURE_ARGS += -DCMAKE_INSTALL_MANDIR="${LOCALBASE}/man" \
+ -DCMAKE_COMPRESS_MAN:BOOL=Off
NO_TEST = Yes
Index: distinfo
===
RCS file: /cvs/ports/graphics/slop/distinfo,v
retrieving revision 1.5
diff -u -p -u -p -r1.5 distinfo
--- distinfo21 Apr 2015 02:07:38 - 1.5
+++ distinfo13 Apr 2017 18:12:20 -
@@ -1,2 +1,2 @@
-SHA256 (slop-4.1.16.tar.gz) = wp8MzbKLxYfJFOmfo/qgmyE129UZp1thSP2kLEFX6Rg=
-SIZE (slop-4.1.16.tar.gz) = 35997
+SHA256 (slop-5.3.37.tar.gz) = OhZW+MzFOrWixv/glYk7Bc9Qo7pdM4V+z0jOP0SAUdw=
+SIZE (slop-5.3.37.tar.gz) = 68182
Index: pkg/PLIST
===
RCS file: /cvs/ports/graphics/slop/pkg/PLIST,v
retrieving revision 1.1.1.1
diff -u -p -u -p -r1.1.1.1 PLIST
--- pkg/PLIST 16 Nov 2014 16:48:56 - 1.1.1.1
+++ pkg/PLIST 13 Apr 2017 18:12:20 -
@@ -1,2 +1,5 @@
@comment $OpenBSD: PLIST,v 1.1.1.1 2014/11/16 16:48:56 bcallah Exp $
@bin bin/slop
+include/slop.hpp
+lib/libslopy.so
Is this a module, or a shared library? The cmake config seems to
specify a library, but doesn't specify a version.
If it should be installed, it should have a version. Something like:
SHARED_LIBS = slopy 0.0 # ?
+@man man/man1/slop.1
--
jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE
