Re: Dovecot (from packages) doesn't work on default OpenBSD 5.2 amd64
On Sun, Feb 03, 2013 at 07:23:36PM -0800, Constantine A. Murenin wrote: > Hi, > > I've installed dovecot-2.1.8 from packages on OpenBSD 5.2 amd64, and > it doesn't seem to be working at all. > > Cns# tail /var/log/maillog > Feb 3 19:13:14 Cns dovecot: master: Error: service(dns_client): > pipe() failed: Too many open files > Feb 3 19:13:14 Cns dovecot: master: Error: service(director): pipe() > failed: Too many open files > Feb 3 19:13:14 Cns dovecot: master: Error: service(dict): pipe() > failed: Too many open files > Feb 3 19:13:14 Cns dovecot: master: Error: service(config): pipe() > failed: Too many open files > Feb 3 19:13:14 Cns dovecot: master: Error: service(auth): pipe() > failed: Too many open files > Feb 3 19:13:14 Cns dovecot: master: Error: service(auth-worker): > pipe() failed: Too many open files > Feb 3 19:13:24 Cns dovecot: master: Error: service(ssl-params): > command startup failed, throttling for 2 secs > Feb 3 19:13:24 Cns dovecot: master: Error: service(anvil): command > startup failed, throttling for 2 secs > Feb 3 19:13:24 Cns dovecot: ssl-params: Fatal: Error reading > configuration: Timeout reading config from /var/dovecot/config > Feb 3 19:13:24 Cns dovecot: anvil: Fatal: Error reading > configuration: Timeout reading config from /var/dovecot/config > Cns# fstat | fgrep -e USER -e dovecot | tail -1 > root dovecot22388 126 pipe 0xfe81d54015b0 state: > Cns# > > It must be hitting the ":openfiles-cur=128" limit from /etc/login.conf. > > What would be the proper way to address this? The package tools point out there is a package readme as soon as the package is installed. -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean.
Dovecot (from packages) doesn't work on default OpenBSD 5.2 amd64
Hi, I've installed dovecot-2.1.8 from packages on OpenBSD 5.2 amd64, and it doesn't seem to be working at all. Cns# tail /var/log/maillog Feb 3 19:13:14 Cns dovecot: master: Error: service(dns_client): pipe() failed: Too many open files Feb 3 19:13:14 Cns dovecot: master: Error: service(director): pipe() failed: Too many open files Feb 3 19:13:14 Cns dovecot: master: Error: service(dict): pipe() failed: Too many open files Feb 3 19:13:14 Cns dovecot: master: Error: service(config): pipe() failed: Too many open files Feb 3 19:13:14 Cns dovecot: master: Error: service(auth): pipe() failed: Too many open files Feb 3 19:13:14 Cns dovecot: master: Error: service(auth-worker): pipe() failed: Too many open files Feb 3 19:13:24 Cns dovecot: master: Error: service(ssl-params): command startup failed, throttling for 2 secs Feb 3 19:13:24 Cns dovecot: master: Error: service(anvil): command startup failed, throttling for 2 secs Feb 3 19:13:24 Cns dovecot: ssl-params: Fatal: Error reading configuration: Timeout reading config from /var/dovecot/config Feb 3 19:13:24 Cns dovecot: anvil: Fatal: Error reading configuration: Timeout reading config from /var/dovecot/config Cns# fstat | fgrep -e USER -e dovecot | tail -1 root dovecot22388 126 pipe 0xfe81d54015b0 state: Cns# It must be hitting the ":openfiles-cur=128" limit from /etc/login.conf. What would be the proper way to address this? C.
sync gcc 4.6 hppa LINK_SPEC
Will soon be needed when ld is changed to deal with static binaries on PIE archs. Index: Makefile === RCS file: /home/pascal/cvs/ports/lang/gcc/4.6/Makefile,v retrieving revision 1.47 diff -u -p -r1.47 Makefile --- Makefile28 Jan 2013 01:03:05 - 1.47 +++ Makefile3 Feb 2013 11:22:24 - @@ -24,14 +24,14 @@ PKGNAME-ada = gnat-${FULL_PKGVERSION} PKGNAME-go = gccgo-${FULL_PKGVERSION} PKGSPEC-main = gcc->=4.6,<4.7 -REVISION-main = 13 -REVISION-c++ = 12 +REVISION-main = 14 +REVISION-c++ = 13 REVISION-estdc = 10 -REVISION-f95 = 12 -REVISION-java = 13 -REVISION-objc = 12 -REVISION-ada = 13 -REVISION-go = 12 +REVISION-f95 = 13 +REVISION-java = 14 +REVISION-objc = 13 +REVISION-ada = 14 +REVISION-go = 13 SHARED_LIBS = estdc++ 14.0 \ gfortran3.0 \ Index: patches/patch-gcc_config_pa_openbsd_h === RCS file: /home/pascal/cvs/ports/lang/gcc/4.6/patches/patch-gcc_config_pa_openbsd_h,v retrieving revision 1.1 diff -u -p -r1.1 patch-gcc_config_pa_openbsd_h --- patches/patch-gcc_config_pa_openbsd_h 16 Jan 2013 19:28:48 - 1.1 +++ patches/patch-gcc_config_pa_openbsd_h 3 Feb 2013 11:18:53 - @@ -1,7 +1,7 @@ $OpenBSD: patch-gcc_config_pa_openbsd_h,v 1.1 2013/01/16 19:28:48 pascal Exp $ gcc/config/pa/openbsd.h.orig Wed Apr 4 05:01:01 2012 -+++ gcc/config/pa/openbsd.hWed Apr 4 08:28:10 2012 -@@ -0,0 +1,186 @@ +--- gcc/config/pa/openbsd.h.orig Sun Feb 3 12:05:20 2013 gcc/config/pa/openbsd.hSun Feb 3 12:18:14 2013 +@@ -0,0 +1,185 @@ +/* Configuration file for an hppa risc OpenBSD target. + + Copyright (C) 1999 Free Software Foundation, Inc. @@ -151,16 +151,15 @@ $OpenBSD: patch-gcc_config_pa_openbsd_h, +#undef TARGET_SCHED_DEFAULT +#define TARGET_SCHED_DEFAULT PROCESSOR_700 + -+/* XXX Why doesn't PA support -R like everyone ??? */ +#undef LINK_SPEC +#define LINK_SPEC \ -+ "%{EB} %{EL} %{shared} %{non_shared} \ -+ %{call_shared} %{no_archive} %{exact_version} \ -+ %{!shared: %{!non_shared: %{!call_shared: -non_shared}}} \ -+ %{!dynamic-linker:-dynamic-linker /usr/libexec/ld.so} \ -+ %{!nostdlib:%{!r*:%{!e*:-e __start}}} -dc -dp \ -+ %{static:-Bstatic} %{!static:-Bdynamic} %{rdynamic:-export-dynamic} \ -+ %{assert*}" ++ "%{!shared:%{!nostdlib:%{!r*:%{!e*:-e __start \ ++ %{shared:-shared} %{R*} \ ++ %{static:-Bstatic} \ ++ %{!static:-Bdynamic} \ ++ %{rdynamic:-export-dynamic} \ ++ %{assert*} \ ++ %{!dynamic-linker:-dynamic-linker /usr/libexec/ld.so}" + +/* Layout of source language data types. */ +
Re: port-modules(5) MODPY_BIN_ADJ
On 2013/02/03 20:26, Stuart Henderson wrote: > MODPY_BIN_ADJ got turned into an internal variable; this reworks > the part of port-modules(5) which referenced it. any comments/ok? rpe points out these are relative to WRKSRC, so tweaked again: Index: port-modules.5 === RCS file: /cvs/src/share/man/man5/port-modules.5,v retrieving revision 1.149 diff -u -p -r1.149 port-modules.5 --- port-modules.5 13 Jan 2013 17:47:21 - 1.149 +++ port-modules.5 3 Feb 2013 20:37:04 - @@ -833,8 +833,7 @@ Sets .Ev MODPY_WANTLIB , .Ev MODPY_LIB_DEPENDS , .Ev MODPY_RUN_DEPENDS , -.Ev MODPY_BUILD_DEPENDS , -.Ev MODPY_BIN_ADJ +.Ev MODPY_BUILD_DEPENDS and .Ev MODPY_ADJ_FILES . Appends to @@ -861,15 +860,11 @@ Extra arguments to the build and install .Ev MODPY_DISTUTILS_BUILDARGS and .Ev MODPY_DISTUTILS_INSTALLARGS . -.Ev MODPY_BIN_ADJ -is a command that takes filename arguments and replaces the python -shebang line with -.Ev MODPY_BIN . -.Ev MODPY_ADJ_FILES -is a list of filenames that will automatically have -.Ev MODPY_BIN_ADJ -called on them at the end of -.Ar pre-configure . +If any files have a python shebang line which needs to be replaced +using MODPY_BIN, list them in +.Ev MODPY_ADJ_FILES . +These are prefixed with WRKSRC and replaced automatically +at the end of .Ar pre-configure . Also affects .Ev CATEGORIES , .Ev MAKE_ENV ,
port-modules(5) MODPY_BIN_ADJ
MODPY_BIN_ADJ got turned into an internal variable; this reworks the part of port-modules(5) which referenced it. any comments/ok? Index: port-modules.5 === RCS file: /cvs/src/share/man/man5/port-modules.5,v retrieving revision 1.149 diff -u -p -r1.149 port-modules.5 --- port-modules.5 13 Jan 2013 17:47:21 - 1.149 +++ port-modules.5 3 Feb 2013 20:25:07 - @@ -833,8 +833,7 @@ Sets .Ev MODPY_WANTLIB , .Ev MODPY_LIB_DEPENDS , .Ev MODPY_RUN_DEPENDS , -.Ev MODPY_BUILD_DEPENDS , -.Ev MODPY_BIN_ADJ +.Ev MODPY_BUILD_DEPENDS and .Ev MODPY_ADJ_FILES . Appends to @@ -861,14 +860,10 @@ Extra arguments to the build and install .Ev MODPY_DISTUTILS_BUILDARGS and .Ev MODPY_DISTUTILS_INSTALLARGS . -.Ev MODPY_BIN_ADJ -is a command that takes filename arguments and replaces the python -shebang line with -.Ev MODPY_BIN . -.Ev MODPY_ADJ_FILES -is a list of filenames that will automatically have -.Ev MODPY_BIN_ADJ -called on them at the end of +If any files have a python shebang line which needs to be replaced +using MODPY_BIN, list them in +.Ev MODPY_ADJ_FILES . +These are replaced automatically at the end of .Ar pre-configure . Also affects .Ev CATEGORIES ,
UPDATE: games/manaplus
Hi ports -- Attached is a diff to update games/manaplus to its latest version. Also taking MAINTAINER, discussed with Jona Joachim (previous MAINTAINER). This works for me on amd64, powerpc, and mips64el (and yes, it's actually *playable* on my yeeloong!) OK? I feel like this should go in before I kill games/tmw, since it's such a significant update. ~Brian ? manaplus13120.diff Index: Makefile === RCS file: /cvs/ports/games/manaplus/Makefile,v retrieving revision 1.3 diff -u -p -r1.3 Makefile --- Makefile 16 Sep 2011 09:48:10 - 1.3 +++ Makefile 3 Feb 2013 17:20:13 - @@ -2,15 +2,14 @@ COMMENT = extended client for The Mana World -V = 1.1.6.26 +V = 1.3.1.20 DISTNAME = manaplus-${V} -REVISION = 1 CATEGORIES = games x11 HOMEPAGE = http://manaplus.evolonline.org/ -MAINTAINER = Jona Joachim +MAINTAINER = Brian Callahan # GPLv2 PERMIT_PACKAGE_CDROM = Yes @@ -23,7 +22,7 @@ WANTLIB += X11 c crypto curl guichan gui WANTLIB += png pthread ssl stdc++ xml2 z MASTER_SITES = http://download.evolonline.org/manaplus/download/${V}/ -EXTRACT_SUFX = .tar.bz2 +EXTRACT_SUFX = .tar.xz MODULES = devel/gettext RUN_DEPENDS= devel/desktop-file-utils @@ -48,8 +47,6 @@ FAKE_FLAGS += desktopdir=${PREFIX}/shar AUTOCONF_VERSION = 2.59 AUTOMAKE_VERSION = 1.9 - -WRKDIST = ${WRKDIR}/manaplus pre-configure: cd ${WRKSRC} && \ Index: distinfo === RCS file: /cvs/ports/games/manaplus/distinfo,v retrieving revision 1.1.1.1 diff -u -p -r1.1.1.1 distinfo --- distinfo 10 Jul 2011 09:56:51 - 1.1.1.1 +++ distinfo 3 Feb 2013 17:20:13 - @@ -1,5 +1,2 @@ -MD5 (manaplus-1.1.6.26.tar.bz2) = h98Pq2knsYKWg3jPVQWcSA== -RMD160 (manaplus-1.1.6.26.tar.bz2) = 5Gq0sTs5hxDvzHKHNOkdL8uHL3Q= -SHA1 (manaplus-1.1.6.26.tar.bz2) = AqiMWJVT5/VfmOEG9u1C8v/jeKQ= -SHA256 (manaplus-1.1.6.26.tar.bz2) = MeJ2sC2VJtoHqfZRCEylO1KpJnzbmpdU6m0jVu4MBag= -SIZE (manaplus-1.1.6.26.tar.bz2) = 4781738 +SHA256 (manaplus-1.3.1.20.tar.xz) = sY5d3DSVCdWYcz441MGFkzN8vFLXZS2Quw+hej648i4= +SIZE (manaplus-1.3.1.20.tar.xz) = 5881004 Index: patches/patch-data_icons_Makefile_am === RCS file: /cvs/ports/games/manaplus/patches/patch-data_icons_Makefile_am,v retrieving revision 1.1.1.1 diff -u -p -r1.1.1.1 patch-data_icons_Makefile_am --- patches/patch-data_icons_Makefile_am 10 Jul 2011 09:56:51 - 1.1.1.1 +++ patches/patch-data_icons_Makefile_am 3 Feb 2013 17:20:14 - @@ -1,6 +1,6 @@ -$OpenBSD: patch-data_icons_Makefile_am,v 1.1.1.1 2011/07/10 09:56:51 jasper Exp $ data/icons/Makefile.am.orig Sat Jun 25 18:12:08 2011 -+++ data/icons/Makefile.am Wed Jun 29 18:46:09 2011 +$OpenBSD$ +--- data/icons/Makefile.am.orig Thu Jan 31 23:27:40 2013 data/icons/Makefile.am Thu Jan 31 23:27:54 2013 @@ -9,8 +9,8 @@ icons_DATA = manaplus.png \ manaplus.xpm Index: patches/patch-src_logger_h === RCS file: patches/patch-src_logger_h diff -N patches/patch-src_logger_h --- /dev/null 1 Jan 1970 00:00:00 - +++ patches/patch-src_logger_h 3 Feb 2013 17:20:14 - @@ -0,0 +1,12 @@ +$OpenBSD$ +--- src/logger.h.orig Thu Jan 31 23:37:09 2013 src/logger.h Thu Jan 31 23:37:20 2013 +@@ -76,7 +76,7 @@ class Logger final + */ + void log(const char *const log_text, ...) + #ifdef __GNUC__ +-__attribute__((__format__(gnu_printf, 2, 3))) ++__attribute__((__format__(printf, 2, 3))) + #endif + ; + Index: patches/patch-src_mumblemanager_cpp === RCS file: /cvs/ports/games/manaplus/patches/patch-src_mumblemanager_cpp,v retrieving revision 1.1.1.1 diff -u -p -r1.1.1.1 patch-src_mumblemanager_cpp --- patches/patch-src_mumblemanager_cpp 10 Jul 2011 09:56:51 - 1.1.1.1 +++ patches/patch-src_mumblemanager_cpp 3 Feb 2013 17:20:14 - @@ -1,7 +1,7 @@ -$OpenBSD: patch-src_mumblemanager_cpp,v 1.1.1.1 2011/07/10 09:56:51 jasper Exp $ src/mumblemanager.cpp.orig Sun Jun 12 13:44:19 2011 -+++ src/mumblemanager.cpp Sun Jun 26 17:46:02 2011 -@@ -7,6 +7,7 @@ +$OpenBSD$ +--- src/mumblemanager.cpp.orig Thu Jan 31 23:28:58 2013 src/mumblemanager.cpp Thu Jan 31 23:29:48 2013 +@@ -16,6 +16,7 @@ #include "utils/mathutils.h" @@ -9,30 +9,3 @@ $OpenBSD: patch-src_mumblemanager_cpp,v #include #include #include -@@ -39,7 +40,7 @@ MumbleManager::~MumbleManager() - { - #ifdef WIN32 - UnmapViewOfFile(mLinkedMem); --#elif defined __FreeBSD__ || defined __DragonFly__ || defined __APPLE__ -+#elif defined BSD4_4 - #else - munmap(mLinkedMem, sizeof(struct LinkedMem)); - #endif -@@ -73,7 +74,7 @@ void MumbleManager::setMapBase(uint16_t mapid) - - void MumbleManager::init() - { --#if defined __FreeBSD__ || defined __DragonFly__ || defined __APPLE__ -+#if de
Re: NEW: Tclthread
On 02/03/13 09:26, Stuart Cassoff wrote: > On 02/03/13 09:02, Stuart Henderson wrote: >> On 2013/02/03 06:58, Stuart Cassoff wrote: >>> On 02/03/13 04:44, Landry Breuil wrote: On Sat, Feb 02, 2013 at 12:38:40PM -0500, Stuart Cassoff wrote: > Provides script-level access to Tcl's threading capabilities. >>> If i get it right it needs tcl sources to build ? Why not making it a subpackage or bundle it with main tcl ? >>> >>> Starting with 8.6, several packages are bundled with the main Tcl >>> distribution. >>> I've split them into separate ports; not all are ready right now. >>> The idea is to use the 8.6 tarball now and change a port to use the package >>> distfile whenever a package is updated. >>> >>> No bundled packages need the Tcl sources to build. >>> The Tcl people (not me) decided with 8.6 to start bundling packages. >>> I had greater interest in getting 8.6 in than complicating things. >>> Some people might like to take all this and make subpackages but I'd >>> prefer to make separate ports and finally, put some notes into the >>> README mentioning the bundled packages and which ports would need >>> to be installed to have a 'full' 8.6+packages installation. >>> A better option would be to depend on tcl/8.6:patch so that you don't need to update tcl version in two places when updating it, and you dont need the extra DISTFILES handling. See for example how mplayer build_depends on ffmpeg:patch. >>> >>> I'll have a look, thanks. >>> >>> >>> Stu >>> (not yet fully caffeinated) >>> >> >> Especially as there will be more ports than just this one, it makes >> sense to sort out the handling of this a bit.. >> >> DISTNAME = thread2.7.0 >> PKGNAME = tclthread-2.7.0 >> DISTFILES = tcl8.6.0-src.tar.gz >> WRKDIST = ${WRKDIR}/tcl8.6.0/pkgs/${DISTNAME} >> >> so... I think DISTNAME should get used for its normal purpose i.e. >> the name of the distfile (DISTFILES is normally just used if you have >> additional files). If not, then at the very least DISTFILES should >> move up with DISTNAME/PKGNAME as otherwise it's more difficult than >> it should be to see where the files are coming from; might be important >> if someone who doesn't know the Tcl ports particularly well if they >> have to update in a hurry e.g. for a security fix. >> >> I presume there will be a lot of commonality between this and the >> ports for other modules, so it would make sense to arrange them in a >> subdirectory so Makefile.inc could be used, or use a module. >> >> I would prefer a subdirectory (maybe lang/tcl86-extensions or >> something..?) to help keep them together rather than stuffing more >> things in the hugely busy devel/ directory. >> >> Hardcoding 8.6.0 in several places seems like a bad idea, I think >> this should come from a module or Makefile.inc somewhere. >> >> >> > > The 8.6 'core' and the bundled packages are all available as separate > tarballs. > I'd like to go that route and forget about the bundled stuff altogether. > The main push behind the bundling was to get TDBC released and 'out there'. > I doubt we'll see (m)any more packages bundled in subsequent releases. > I don't know what to do about directory overcrowding, not really my bailiwick. > Here's version of the port that uses its own distfile instead of the tcl-bundled one. Stu tclthread-2.7.0-port.tar.gz Description: application/gzip
Re: NEW: Tcl/Tk 8.6.0
On 01/14/13 07:12, Stuart Cassoff wrote: > Plus the patches for tcl.port.mk and tk.port.mk. > Tested mostly on i386 and somewhat on amd64. Use the core dist instead of the bundled one. Index: Makefile === RCS file: /cvs/ports/lang/tcl/8.6/Makefile,v retrieving revision 1.1.1.1 diff -u -p -r1.1.1.1 Makefile --- Makefile2 Feb 2013 11:15:01 - 1.1.1.1 +++ Makefile3 Feb 2013 16:00:21 - @@ -4,6 +4,7 @@ COMMENT = Tool Command Language DISTNAME = tcl8.6.0 PKGNAME = tcl-8.6.0 +REVISION = 0 SHARED_LIBS = tcl86 0.0 CATEGORIES = lang lang/tcl HOMEPAGE = http://www.tcl.tk/ @@ -18,7 +19,7 @@ PERMIT_DISTFILES_FTP =Yes WANTLIB = c m pthread z MASTER_SITES = ${MASTER_SITE_SOURCEFORGE:=tcl/} -DISTFILES =${DISTNAME}-src.tar.gz +DISTFILES =tcl-core8.6.0-src.tar.gz MAKE_FLAGS += TCL_PACKAGE_PATH='$$(prefix)/lib/tcl' \ TCL_LIBRARY='$$(prefix)/lib/tcl/tcl$$(VERSION)' Index: distinfo === RCS file: /cvs/ports/lang/tcl/8.6/distinfo,v retrieving revision 1.1.1.1 diff -u -p -r1.1.1.1 distinfo --- distinfo2 Feb 2013 11:15:01 - 1.1.1.1 +++ distinfo3 Feb 2013 16:00:21 - @@ -1,2 +1,2 @@ -SHA256 (tcl8.6.0-src.tar.gz) = NUQiucR5FoVJkSOy3+Afqpi1VcCJBsAQy0RJ3cddyt4= -SIZE (tcl8.6.0-src.tar.gz) = 8636908 +SHA256 (tcl-core8.6.0-src.tar.gz) = xTkVmWuz3doF12X4SLvVURHUS3RAaptnoCBwB4pkWWQ= +SIZE (tcl-core8.6.0-src.tar.gz) = 5719946
Re: opensc-pkcs11.so
Stuart, thank you. It works. On 14:15 Sun 03 Feb , Stuart Henderson wrote: > On 2013/02/03 17:57, Sergey Bronnikov wrote: > > Hi, everyone > > > > may be someone knows why attempt to load that module failed: > > > > ~$ ssh -I /usr/local/lib/pkcs11/opensc-pkcs11.so gw > > dlopen /usr/local/lib/pkcs11/opensc-pkcs11.so failed: Cannot load specified > > object > > ssh isn't linked with libpthread, so the symbols required by > opensc-pkcs11.so aren't available. > > The issue is that various weak-aliased functions in libc are overridden > by functions in libpthread. If a shared library/module requires libpthread > functions, it can only be dlopen()'d at runtime if libpthread was preloaded > or was specifically linked to the executable. > > The normal library dependency resolution has been specifically blocked > for libpthread to ensure that the functions get overridden. > > Try preloading libpthread like this: > > $ LD_PRELOAD=/usr/lib/libpthread.so ssh \ > -I /usr/local/lib/pkcs11/opensc-pkcs11.so gw > > > > > ~$ pkcs11-tool -I > > > > error: Failed to load pkcs11 module > > > > Aborting. > > Same problem, and the LD_PRELOAD hack should work too, though in this > case it would probably be better if pkcs11-tool (and any of the other > binaries provided by that package) were linked with -lpthread. > > > On 10:39 Sun 03 Feb , Douglas Santos wrote: > > > I'm not actively maintaining this port anymore. You can try to ask on > > > ports@openbsd > > I have dropped the MAINTAINER entry from the port. > -- sergeyb@
Re: NEW: Tclthread
On 02/03/13 09:02, Stuart Henderson wrote: > On 2013/02/03 06:58, Stuart Cassoff wrote: >> On 02/03/13 04:44, Landry Breuil wrote: >>> On Sat, Feb 02, 2013 at 12:38:40PM -0500, Stuart Cassoff wrote: Provides script-level access to Tcl's threading capabilities. >>> >> >>> If i get it right it needs tcl sources to build ? Why not making it a >>> subpackage or bundle it with main tcl ? >> >> Starting with 8.6, several packages are bundled with the main Tcl >> distribution. >> I've split them into separate ports; not all are ready right now. >> The idea is to use the 8.6 tarball now and change a port to use the package >> distfile whenever a package is updated. >> >> No bundled packages need the Tcl sources to build. >> The Tcl people (not me) decided with 8.6 to start bundling packages. >> I had greater interest in getting 8.6 in than complicating things. >> Some people might like to take all this and make subpackages but I'd >> prefer to make separate ports and finally, put some notes into the >> README mentioning the bundled packages and which ports would need >> to be installed to have a 'full' 8.6+packages installation. >> >>> A better option would be to depend on tcl/8.6:patch so that you don't >>> need to update tcl version in two places when updating it, and you dont >>> need the extra DISTFILES handling. See for example how mplayer >>> build_depends on ffmpeg:patch. >> >> I'll have a look, thanks. >> >> >> Stu >> (not yet fully caffeinated) >> > > Especially as there will be more ports than just this one, it makes > sense to sort out the handling of this a bit.. > > DISTNAME = thread2.7.0 > PKGNAME = tclthread-2.7.0 > DISTFILES = tcl8.6.0-src.tar.gz > WRKDIST = ${WRKDIR}/tcl8.6.0/pkgs/${DISTNAME} > > so... I think DISTNAME should get used for its normal purpose i.e. > the name of the distfile (DISTFILES is normally just used if you have > additional files). If not, then at the very least DISTFILES should > move up with DISTNAME/PKGNAME as otherwise it's more difficult than > it should be to see where the files are coming from; might be important > if someone who doesn't know the Tcl ports particularly well if they > have to update in a hurry e.g. for a security fix. > > I presume there will be a lot of commonality between this and the > ports for other modules, so it would make sense to arrange them in a > subdirectory so Makefile.inc could be used, or use a module. > > I would prefer a subdirectory (maybe lang/tcl86-extensions or > something..?) to help keep them together rather than stuffing more > things in the hugely busy devel/ directory. > > Hardcoding 8.6.0 in several places seems like a bad idea, I think > this should come from a module or Makefile.inc somewhere. > > > The 8.6 'core' and the bundled packages are all available as separate tarballs. I'd like to go that route and forget about the bundled stuff altogether. The main push behind the bundling was to get TDBC released and 'out there'. I doubt we'll see (m)any more packages bundled in subsequent releases. I don't know what to do about directory overcrowding, not really my bailiwick. Stu
Re: opensc-pkcs11.so
On 2013/02/03 17:57, Sergey Bronnikov wrote: > Hi, everyone > > may be someone knows why attempt to load that module failed: > > ~$ ssh -I /usr/local/lib/pkcs11/opensc-pkcs11.so gw > dlopen /usr/local/lib/pkcs11/opensc-pkcs11.so failed: Cannot load specified > object ssh isn't linked with libpthread, so the symbols required by opensc-pkcs11.so aren't available. The issue is that various weak-aliased functions in libc are overridden by functions in libpthread. If a shared library/module requires libpthread functions, it can only be dlopen()'d at runtime if libpthread was preloaded or was specifically linked to the executable. The normal library dependency resolution has been specifically blocked for libpthread to ensure that the functions get overridden. Try preloading libpthread like this: $ LD_PRELOAD=/usr/lib/libpthread.so ssh \ -I /usr/local/lib/pkcs11/opensc-pkcs11.so gw > > > ~$ pkcs11-tool -I > > > error: Failed to load pkcs11 module > > > Aborting. Same problem, and the LD_PRELOAD hack should work too, though in this case it would probably be better if pkcs11-tool (and any of the other binaries provided by that package) were linked with -lpthread. > On 10:39 Sun 03 Feb , Douglas Santos wrote: > > I'm not actively maintaining this port anymore. You can try to ask on > > ports@openbsd I have dropped the MAINTAINER entry from the port.
Re: NEW: Tclthread
On 2013/02/03 06:58, Stuart Cassoff wrote: > On 02/03/13 04:44, Landry Breuil wrote: > > On Sat, Feb 02, 2013 at 12:38:40PM -0500, Stuart Cassoff wrote: > >> Provides script-level access to Tcl's threading capabilities. > > > > > If i get it right it needs tcl sources to build ? Why not making it a > > subpackage or bundle it with main tcl ? > > Starting with 8.6, several packages are bundled with the main Tcl > distribution. > I've split them into separate ports; not all are ready right now. > The idea is to use the 8.6 tarball now and change a port to use the package > distfile whenever a package is updated. > > No bundled packages need the Tcl sources to build. > The Tcl people (not me) decided with 8.6 to start bundling packages. > I had greater interest in getting 8.6 in than complicating things. > Some people might like to take all this and make subpackages but I'd > prefer to make separate ports and finally, put some notes into the > README mentioning the bundled packages and which ports would need > to be installed to have a 'full' 8.6+packages installation. > > > A better option would be to depend on tcl/8.6:patch so that you don't > > need to update tcl version in two places when updating it, and you dont > > need the extra DISTFILES handling. See for example how mplayer > > build_depends on ffmpeg:patch. > > I'll have a look, thanks. > > > Stu > (not yet fully caffeinated) > Especially as there will be more ports than just this one, it makes sense to sort out the handling of this a bit.. DISTNAME = thread2.7.0 PKGNAME = tclthread-2.7.0 DISTFILES = tcl8.6.0-src.tar.gz WRKDIST = ${WRKDIR}/tcl8.6.0/pkgs/${DISTNAME} so... I think DISTNAME should get used for its normal purpose i.e. the name of the distfile (DISTFILES is normally just used if you have additional files). If not, then at the very least DISTFILES should move up with DISTNAME/PKGNAME as otherwise it's more difficult than it should be to see where the files are coming from; might be important if someone who doesn't know the Tcl ports particularly well if they have to update in a hurry e.g. for a security fix. I presume there will be a lot of commonality between this and the ports for other modules, so it would make sense to arrange them in a subdirectory so Makefile.inc could be used, or use a module. I would prefer a subdirectory (maybe lang/tcl86-extensions or something..?) to help keep them together rather than stuffing more things in the hugely busy devel/ directory. Hardcoding 8.6.0 in several places seems like a bad idea, I think this should come from a module or Makefile.inc somewhere.
Re: opensc-pkcs11.so
Hi, everyone may be someone knows why attempt to load that module failed: ~$ ssh -I /usr/local/lib/pkcs11/opensc-pkcs11.so gw dlopen /usr/local/lib/pkcs11/opensc-pkcs11.so failed: Cannot load specified object On 10:39 Sun 03 Feb , Douglas Santos wrote: > Hi there, > sorry for the late. > I'm not actively maintaining this port anymore. You can try to ask on > ports@openbsd > > On Mon, Jan 28, 2013 at 5:05 AM, Sergey Bronnikov wrote: > > > > Hi, Douglas. > > > > > > Did you test ssh client with retrive SSH keys from token? > > > > ~$ ssh -I /usr/local/lib/pkcs11/opensc-pkcs11.so gw > > dlopen /usr/local/lib/pkcs11/opensc-pkcs11.so failed: Cannot load specified > > object > > > > ~$ pkcs11-tool -I > > error: Failed to load pkcs11 module > > Aborting. > > > > while opensc successfully detects my token: > > > > /usr/ports$ opensc-tool -n > > Using reader with a card: Rutoken S driver > > Rutoken S card > > /usr/ports$ opensc-tool -a > > Using reader with a card: Rutoken S driver > > ... > > > > I use latest OpenBSD snapshot, amd64. > > > > -- > > sergeyb@ -- sergeyb@
Re: Remove games/tmw (was: UPDATE: devel/guichan)
if upstream killed it, then please kill it. On 2013 Feb 02 (Sat) at 11:05:15 -0500 (-0500), Brian Callahan wrote: :Hi ports -- : :As I mentioned in the UPDATE: devel/guichan thread, the mana client :found in games/tmw is being deprecated by upstream. They have blessed :the manaplus client (games/manaplus) as the official client. Upstream :claims the most recent version of the mana client (which is not the :version we have in ports) will be supported through October 2013 but :any new features may not be available. :From upstream: :"In fact, since there will likely be no more releases of the Mana :client that supports TMWA (this server), there will inevitably be a :time when ManaPlus becomes the only supported client." : :Since I use the manaplus client and the writing is on the wall for :the mana client, I'd prefer to remove games/tmw now since the :alternative is to update the port, support it until October, and then :have this conversation again. : :Any objections? : :~Brian : -- Man 1: Ask me what the most important thing about telling a good joke is. Man 2: OK, what is the most impo -- Man 1: __TIMING!
UPDATE exmh to use Tcl/Tk 8.5.
Fonts are better but a little smaller than when running with 8.4. This can be adjusted with the -size option in the patch. Stu Index: Makefile === RCS file: /cvs/ports/mail/exmh2/Makefile,v retrieving revision 1.31 diff -u -p -r1.31 Makefile --- Makefile19 Nov 2010 07:23:06 - 1.31 +++ Makefile3 Feb 2013 12:07:27 - @@ -3,7 +3,7 @@ COMMENT= Tcl/Tk interface to the MH mail system DISTNAME= exmh-2.7.2 -REVISION = 0 +REVISION = 1 CATEGORIES=mail HOMEPAGE= http://www.beedub.com/exmh/ @@ -20,9 +20,6 @@ MASTER_SITES= ftp://ftp.funet.fi/pub/la ftp://ftp.tcl.tk/pub/tcl/exmh/ MODULES= x11/tk -# XXX broken fonts with 8.5 -MODTCL_VERSION=8.4 -MODTK_VERSION= 8.4 BUILD_DEPENDS= ${MODTCL_RUN_DEPENDS} RUN_DEPENDS= lang/expect\ mail/nmh\ Index: distinfo === RCS file: /cvs/ports/mail/exmh2/distinfo,v retrieving revision 1.6 diff -u -p -r1.6 distinfo --- distinfo5 Apr 2007 16:20:01 - 1.6 +++ distinfo3 Feb 2013 12:07:27 - @@ -1,5 +1,2 @@ -MD5 (exmh-2.7.2.tar.gz) = /bfG/ybQQp6pUFkKNvE2nw== -RMD160 (exmh-2.7.2.tar.gz) = /Q2GiTBM223LzTQexbpfj1svlxI= -SHA1 (exmh-2.7.2.tar.gz) = Do4Rwpx+G5Nx1acrTLK/YIXyDt8= SHA256 (exmh-2.7.2.tar.gz) = 9FM5UL1shDQn/K4o4JRkVMlOtHdF8gJWuEfbpUC8Xz4= SIZE (exmh-2.7.2.tar.gz) = 1081601 Index: files/configure === RCS file: /cvs/ports/mail/exmh2/files/configure,v retrieving revision 1.2 diff -u -p -r1.2 configure --- files/configure 13 Mar 2009 15:17:02 - 1.2 +++ files/configure 3 Feb 2013 12:07:27 - @@ -12,8 +12,8 @@ my $domain = ""; my $prefix = exists $ENV{"PREFIX"} ? $ENV{"PREFIX"} : "/usr/local"; my $localbase = exists $ENV{"LOCALBASE"} ? $ENV{"LOCALBASE"} : "/usr/local"; -my $modtcl_bin = exists $ENV{"MODTCL_BIN"} ? $ENV{"MODTCL_BIN"} : "$localbase/bin/tclsh.4"; -my $modtk_bin = exists $ENV{"MODTK_BIN"} ? $ENV{"MODTK_BIN"} : "$localbase/bin/wish8.4"; +my $modtcl_bin = exists $ENV{"MODTCL_BIN"} ? $ENV{"MODTCL_BIN"} : "$localbase/bin/tclsh8.5"; +my $modtk_bin = exists $ENV{"MODTK_BIN"} ? $ENV{"MODTK_BIN"} : "$localbase/bin/wish8.5"; my $config = " # Saved state from exmh.install # Mon Mar 14 10:59:18 MST 2005 Index: patches/patch-exmh_MASTER === RCS file: patches/patch-exmh_MASTER diff -N patches/patch-exmh_MASTER --- /dev/null 1 Jan 1970 00:00:00 - +++ patches/patch-exmh_MASTER 3 Feb 2013 12:07:27 - @@ -0,0 +1,14 @@ +$OpenBSD$ +--- exmh.MASTER.orig Mon Jun 14 11:24:58 2004 exmh.MASTERSat Feb 2 18:05:56 2013 +@@ -28,6 +28,10 @@ if {$tcl_version < 8.0} { + # appear all alone. + wm withdraw . + ++if {[package vsatisfies [package present Tcl] 8.5]} { ++font create fixed {*}[dict merge [font configure TkFixedFont] [dict create -size -10]] ++} ++ + proc auto_path_update { path } { + # Add library directories to the auto_path, + # ensuring that later paths have precedence
Re: NEW: Tclthread
On 02/03/13 04:44, Landry Breuil wrote: > On Sat, Feb 02, 2013 at 12:38:40PM -0500, Stuart Cassoff wrote: >> Provides script-level access to Tcl's threading capabilities. > > If i get it right it needs tcl sources to build ? Why not making it a > subpackage or bundle it with main tcl ? Starting with 8.6, several packages are bundled with the main Tcl distribution. I've split them into separate ports; not all are ready right now. The idea is to use the 8.6 tarball now and change a port to use the package distfile whenever a package is updated. No bundled packages need the Tcl sources to build. The Tcl people (not me) decided with 8.6 to start bundling packages. I had greater interest in getting 8.6 in than complicating things. Some people might like to take all this and make subpackages but I'd prefer to make separate ports and finally, put some notes into the README mentioning the bundled packages and which ports would need to be installed to have a 'full' 8.6+packages installation. > A better option would be to depend on tcl/8.6:patch so that you don't > need to update tcl version in two places when updating it, and you dont > need the extra DISTFILES handling. See for example how mplayer > build_depends on ffmpeg:patch. I'll have a look, thanks. Stu (not yet fully caffeinated)
sox 14.4.1
A maintainance release of sox has been released. Tested on current/i386 and current/amd64. Jan Index: Makefile === RCS file: /cvs/ports/audio/sox/Makefile,v retrieving revision 1.53 diff -u -p -r1.53 Makefile --- Makefile9 Jul 2012 13:25:37 - 1.53 +++ Makefile3 Feb 2013 12:17:17 - @@ -2,13 +2,12 @@ COMMENT= Sound eXchange, the Swiss Army knife of audio manipulation -DISTNAME= sox-14.4.0 -SHARED_LIBS += sox 3.0 # .2.0 +DISTNAME= sox-14.4.1 +SHARED_LIBS += sox 3.0 # .2.1 CATEGORIES=audio HOMEPAGE= http://sox.sourceforge.net/ MAINTAINER=Jan Stary -REVISION= 1 # code is GPLv2+, however as opencore is under Apache License we must use # sox under the terms of GPLv3 to avoid license conflict. Index: distinfo === RCS file: /cvs/ports/audio/sox/distinfo,v retrieving revision 1.12 diff -u -p -r1.12 distinfo --- distinfo23 Mar 2012 13:16:41 - 1.12 +++ distinfo3 Feb 2013 12:17:17 - @@ -1,5 +1,2 @@ -MD5 (sox-14.4.0.tar.gz) = sMFc/3pLoOwX/cdOah+c8Q== -RMD160 (sox-14.4.0.tar.gz) = rxtsAtKcmI+bDf7Jyp67MJSbBmM= -SHA1 (sox-14.4.0.tar.gz) = 2AnKs4LHqdAVSRxpBRqdHBoaRPE= -SHA256 (sox-14.4.0.tar.gz) = PuNLFN0mfeN46KEXqugexMrjMHcjQualW79lIKCoiqM= -SIZE (sox-14.4.0.tar.gz) = 1105756 +SHA256 (sox-14.4.1.tar.gz) = mowsb+UeYI2jRqFX4RFQipV6+ePs894meB026aZ/qJs= +SIZE (sox-14.4.1.tar.gz) = 653
Re: UPDATE: graphics/cairo
- Original message - > On 2013/02/03 09:57, Landry Breuil wrote: > > On Sat, Feb 02, 2013 at 06:21:49AM -0500, Brad Smith wrote: > > > On Fri, Feb 01, 2013 at 07:27:22PM +0100, Eric Faurot wrote: > > > > Hi, > > > > > > > > This is an update to cairo 1.12.12. > > > > > > > > Eric. > > > > > > Here is a better diff. > > > > It's not a better diff until you've said what it changes compared to > > eric's diff. Why adding a dep on lzo2 ? > > Cairo uses lzo2 now, check the changelog..also fixes a hardcoded > path to x11. This is why. > Diff looks good but I am a little twitchy about updating a widely used > library at this point, bearing in mind that if it causes problems there > might not be time to fix them, this at least needs some wide testing. > It's working ok for what I use on amd64 but could do with tests on > more arches.. I think this should be good for the release with the bug fixes it has but yes it should be tested a fair bit. -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean.
Re: NEW: Tclthread
On Sat, Feb 02, 2013 at 12:38:40PM -0500, Stuart Cassoff wrote: > Provides script-level access to Tcl's threading capabilities. If i get it right it needs tcl sources to build ? Why not making it a subpackage or bundle it with main tcl ? A better option would be to depend on tcl/8.6:patch so that you don't need to update tcl version in two places when updating it, and you dont need the extra DISTFILES handling. See for example how mplayer build_depends on ffmpeg:patch. Landry
Re: UPDATE: graphics/cairo
On 2013/02/03 09:57, Landry Breuil wrote: > On Sat, Feb 02, 2013 at 06:21:49AM -0500, Brad Smith wrote: > > On Fri, Feb 01, 2013 at 07:27:22PM +0100, Eric Faurot wrote: > > > Hi, > > > > > > This is an update to cairo 1.12.12. > > > > > > Eric. > > > > Here is a better diff. > > It's not a better diff until you've said what it changes compared to > eric's diff. Why adding a dep on lzo2 ? Cairo uses lzo2 now, check the changelog..also fixes a hardcoded path to x11. Diff looks good but I am a little twitchy about updating a widely used library at this point, bearing in mind that if it causes problems there might not be time to fix them, this at least needs some wide testing. It's working ok for what I use on amd64 but could do with tests on more arches..
Re: extend graphics/DevIL
On Wed, January 30, 2013 16:11, Kirill Bychkov wrote: > On Wed, January 30, 2013 15:58, David Coppa wrote: >> On Wed, Jan 30, 2013 at 1:55 PM, Kirill Bychkov wrote: >>> Hi. This patch extends functionality of graphics/devil: >>> - enables ILU and ILUT libraries >>> - add ilur utility for graphics transform in command line >>> >>> Regen distinfo while here. >>> ilur tested on amd64 and works fine. OK to commit? >> >> Since you're here, could you get rid of PFRAG.shared? > > Done ping? > Index: Makefile > === > RCS file: /cvs/ports/graphics/DevIL/Makefile,v > retrieving revision 1.8 > diff -u -p -r1.8 Makefile > --- Makefile 20 Sep 2012 14:04:17 - 1.8 > +++ Makefile 30 Jan 2013 13:09:25 - > @@ -3,8 +3,10 @@ > COMMENT= library for powerful image loading capabilities > > DISTNAME=DevIL-1.7.8 > -REVISION=5 > +REVISION=6 > SHARED_LIBS+=IL 0.0 # 2.0 > +SHARED_LIBS+=ILU 0.0 # 2.0 > +SHARED_LIBS+=ILUT0.0 # 2.0 > > CATEGORIES= graphics devel > > @@ -19,10 +21,15 @@ PERMIT_DISTFILES_FTP= Yes > MASTER_SITES=${MASTER_SITE_SOURCEFORGE:=openil/} > EXTRACT_SUFX=.zip > > -WANTLIB += Half Iex IlmImf IlmThread Imath jasper jpeg lcms m > -WANTLIB += mng png pthread stdc++ tiff z > - > -LIB_DEPENDS= graphics/ilmbase \ > +WANTLIB += GL GLU Half ICE Iex IlmImf IlmThread Imath SDL SM X11 > +WANTLIB += Xdamage Xext Xfixes Xi Xmu Xrandr Xrender Xt Xxf86vm > +WANTLIB += c drm glut jasper jpeg lcms m mng png pthread sndio > +WANTLIB += stdc++ tiff usbhid xcb z > + > +MODULES= converters/libiconv > +LIB_DEPENDS= devel/sdl \ > + graphics/freeglut \ > + graphics/ilmbase \ > graphics/jasper \ > graphics/libmng \ > graphics/lcms \ > @@ -36,7 +43,10 @@ USE_LIBTOOL= Yes > AUTOCONF_VERSION=2.61 > CONFIGURE_STYLE= autoconf > CONFIGURE_ARGS+= --with-squish=no \ > - --with-zlib=yes > + --with-zlib=yes \ > + --enable-ILU \ > +--enable-ILUT > + > # This would only install three binary examples, we'll install it manually. > CONFIGURE_ARGS+= --with-examples=no > CONFIGURE_ENV= CPPFLAGS="-I${LOCALBASE}/include > -I${LOCALBASE}/include/libpng/ -I${X11BASE}/include" \ > Index: distinfo > === > RCS file: /cvs/ports/graphics/DevIL/distinfo,v > retrieving revision 1.1.1.1 > diff -u -p -r1.1.1.1 distinfo > --- distinfo 30 Mar 2011 09:27:58 - 1.1.1.1 > +++ distinfo 30 Jan 2013 13:09:25 - > @@ -1,5 +1,2 @@ > -MD5 (DevIL-1.7.8.zip) = MShT75yFrXshAPnNBosqWw== > -RMD160 (DevIL-1.7.8.zip) = uSiKTlPpoGEPBLS7H859hf2EhK8= > -SHA1 (DevIL-1.7.8.zip) = V4fhY12Pqpvxrsjeh83TlhTXLJc= > SHA256 (DevIL-1.7.8.zip) = Q2jYOzAWter+iYTw1/hszudriZPPkA06jL1JAbUvZOs= > SIZE (DevIL-1.7.8.zip) = 3452549 > Index: patches/patch-src-ILUT_src_ilut_opengl_c > === > RCS file: patches/patch-src-ILUT_src_ilut_opengl_c > diff -N patches/patch-src-ILUT_src_ilut_opengl_c > --- /dev/null 1 Jan 1970 00:00:00 - > +++ patches/patch-src-ILUT_src_ilut_opengl_c 30 Jan 2013 13:09:25 - > @@ -0,0 +1,48 @@ > +$OpenBSD$ > +--- src-ILUT/src/ilut_opengl.c.orig Mon Sep 14 15:22:33 2009 > src-ILUT/src/ilut_opengl.c Mon Sep 14 15:25:13 2009 > +@@ -58,7 +58,7 @@ void *aglGetProcAddress( const GLubyte *name ) { > + // #pragma comment(lib, "freeglut.lib") > + #endif > + > +-#ifdef linux > ++#if defined(linux) || defined (__OpenBSD__) > + // fix for glXGetProcAddressARB > + #define GLX_GLXEXT_PROTOTYPES > + #include > +@@ -84,7 +84,7 @@ void *aglGetProcAddress( const GLubyte *name ) { > + > + static ILboolean HasCubemapHardware = IL_FALSE; > + static ILboolean HasNonPowerOfTwoHardware = IL_FALSE; > +-#if defined(_WIN32) || defined(_WIN64) || defined(linux) || > defined(__APPLE__) > ++#if defined(_WIN32) || defined(_WIN64) || defined(linux) || > defined(__APPLE__) || defined (__OpenBSD__) > + ILGLTEXIMAGE3DARBPROC ilGLTexImage3D = NULL; > + ILGLTEXSUBIMAGE3DARBPROCilGLTexSubImage3D = NULL; > + ILGLCOMPRESSEDTEXIMAGE2DARBPROC ilGLCompressed2D = NULL; > +@@ -127,7 +127,7 @@ ILboolean ilutGLInit() > + IsExtensionSupported("GL_EXT_texture3D")) { > + ilGLCompressed3D = > (ILGLCOMPRESSEDTEXIMAGE3DARBPROC)wglGetProcAddress("glCompressedTexImage3DARB"); > + } > +-#elif linux > ++#elif defined(linux) || defined (__OpenBSD__) > + if (IsExtensionSupported("GL_ARB_texture_compression") && > + > IsExtensionSuppor
Re: UPDATE: graphics/cairo
On Sat, Feb 02, 2013 at 06:21:49AM -0500, Brad Smith wrote: > On Fri, Feb 01, 2013 at 07:27:22PM +0100, Eric Faurot wrote: > > Hi, > > > > This is an update to cairo 1.12.12. > > > > Eric. > > Here is a better diff. It's not a better diff until you've said what it changes compared to eric's diff. Why adding a dep on lzo2 ? Landry