Re: [update] sysutils/entr 4.3 -> 4.4
On Sun, Feb 02, 2020 at 10:04:00PM +0100, Charlene Wendling wrote: > Hi, > > Here is a quick update for sysutils/entr. > > According to $WRKSRC/NEWS, the only change impacting us is the > introduction of the "ENTR_INOTIFY_WORKAROUND" environment variable > to enable a compatibility mode for platforms with "deformed" inotify > support. All of the significant changes in entr-4.4 are changes to INOTIFY support, there are no changes to functionality on BSD. > Port-wise, i've moved HOMEPAGE and MASTER_SITES to the new https URL, > the previous one contains only a link to the new one. I added https to the project web site because people kept asking me for it. As far as ports go, do we consider https:// to be better? But please do update the HOMEPAGE location.
Re: UPDATE: sysutils/entr 4.0 -> 4.1
On Mon, Dec 31, 2018 at 10:55:14PM +0100, Sascha Paunovic wrote: > Hi, > > here an update of sysutils/entr to 4.1. Tested on amd64, works ok for me. Yes, this update is good. > Index: Makefile > === > RCS file: /cvs/ports/sysutils/entr/Makefile,v > retrieving revision 1.12 > diff -u -p -r1.12 Makefile > --- Makefile 20 Feb 2018 19:23:34 - 1.12 > +++ Makefile 31 Dec 2018 21:53:23 - > @@ -2,8 +2,8 @@ > > COMMENT =run arbitrary commands when files change > > -DISTNAME = entr-4.0 > -REV =d5110481f5b9 > +DISTNAME = entr-4.1 > +REV =f4e2cbe57708 > > CATEGORIES = sysutils > > Index: distinfo > === > RCS file: /cvs/ports/sysutils/entr/distinfo,v > retrieving revision 1.9 > diff -u -p -r1.9 distinfo > --- distinfo 20 Feb 2018 19:23:34 - 1.9 > +++ distinfo 31 Dec 2018 21:53:23 - > @@ -1,2 +1,2 @@ > -SHA256 (entr-4.0.tar.gz) = StT+kQixeRmZUc/HilgaimlgKwc9rlm8rkuBD24fbIs= > -SIZE (entr-4.0.tar.gz) = 24758 > +SHA256 (entr-4.1.tar.gz) = xQPJPM/8HmstqXnNy4jFpobiYRA1Ae7ecHf6CJzf73g= > +SIZE (entr-4.1.tar.gz) = 25265
Re: [UPDATE] sysutils/entr 3.7 -> 3.9
On Sun, Nov 12, 2017 at 03:12:21AM -0500, Brian Callahan wrote: > > On 11/11/17 22:58, Klemens Nanni wrote: > > Straight forward update, all tests pass and entr continues to work fine > > on my amd64 machine. > > > > Feedback? Does anyone want to commit this? > > Did you reach out to the MAINTAINER? What did he say? I happen to know that > the MAINTAINER is also upstream for this port. > > I've CC'd him here. I will commit your diff after I hear back from him This is good. entr-3.9 has been a solid release. (And I run OpenBSD on my workstation so it should be unusual to see a regression on this particular platform!) > > > diff --git a/sysutils/entr/Makefile b/sysutils/entr/Makefile > > index 49bdd79fa1d..ca6691145c1 100644 > > --- a/sysutils/entr/Makefile > > +++ b/sysutils/entr/Makefile > > @@ -2,8 +2,8 @@ > > COMMENT = run arbitrary commands when files change > > -DISTNAME = entr-3.7 > > -REV = c5b62bde107d > > +DISTNAME = entr-3.9 > > +REV = 332fd96a324a > > CATEGORIES = sysutils > > diff --git a/sysutils/entr/distinfo b/sysutils/entr/distinfo > > index bffb20b029b..1bbea369779 100644 > > --- a/sysutils/entr/distinfo > > +++ b/sysutils/entr/distinfo > > @@ -1,2 +1,2 @@ > > -SHA256 (entr-3.7.tar.gz) = lO/VDI9+nVaQYNXe6/Nmw1ZegegUqzMrlz1ymPqOoi8= > > -SIZE (entr-3.7.tar.gz) = 23326 > > +SHA256 (entr-3.9.tar.gz) = AtePGK5TDmS/u52OAlCWL4WUbhCFDdBliZ0DrxXyaHY= > > +SIZE (entr-3.9.tar.gz) = 24554 > >
Re: PostgreSQL 9.4 RC1
On Wed, Dec 17, 2014 at 02:28:15PM +0100, Pierre-Emmanuel Andr? wrote: On Wed, Dec 17, 2014 at 08:01:45AM -0500, Eric Radman wrote: On Tue, Nov 25, 2014 at 02:33:10PM +0100, Pierre-Emmanuel Andr? wrote: Hi, If you want to test the new PostgreSQL, there is a diff for the 9.4 RC1. Tested on @amd64 Regards, In testing environments 9.4rc1 has been running well so far, but one application I run broke because the uuid-ossp extension is not included in the contib package. What is the motive for disabling uuid-ossp? Nice, thanks for your report. As mentionned by jeremy@, portswise, ossp-uuid is no longer used, as pgcrypto now has built in support for creation of version 4 UUIDs. I missed that. You are right; this is all that is required for me to adapt to 9.4: -CREATE EXTENSION uuid-ossp; +CREATE EXTENSION pgcrypto; -id uuid PRIMARY KEY default uuid_generate_v4(), +id uuid PRIMARY KEY default gen_random_uuid(), Could you tell me which application needs it ? It's an internal app that was easily adapted. Thanks again.
Re: Update: phantomjs 1.9.8
On Sat, Nov 01, 2014 at 10:23:01AM +0100, Francisco de Borja Lopez Rio wrote: Hi! Attached is a diff to update the www/phantomjs port to the latest version (1.9.8) released some days ago. This version disables SSLv3 support in phantomjs because of CVE-2014-3566 (aka the POODLE issue). Not much more changes since 1.9.7: https://github.com/ariya/phantomjs/compare/1.9.7...1.9.8 I've built the packages without problem in both latest amd64 snapshot and an i386 snapshot from the end of may. It would be nice if someone with an up-to-date i386 snapshot could give the build a try. Some light testing using Selenium seems to indicate that this update works fine on i386. Tested using an October 21 snapshot.
Re: rc_script questions.
On Tue, Aug 26, 2014 at 11:37:49AM -0500, Chuck Burns wrote: It -does- exit the script, and the rc_start() command is -not- processed.. so I know that part is working.. However, there is no output, and the script doesn't fail it just quits, silently... Try running the rc script with the -d option. Normally stdin/stdout is redirected to /dev/null (see rc.d(8))
Re: py-selenium and OpenBSD
On Thu, Jun 19, 2014 at 08:21:11PM -0700, Craig Paulette wrote: I'm attempting to do some web development on OpenBSD but can't get py-selenium to work. This thread shows that I'm not alone: http://openbsd.7691.n7.nabble.com/py-selenium-td238035.html I can't be the only one trying to use py-selenium on OpenBSD. What is everyone else doing as a workaround in the meantime? What browser were you trying to use with py-selenium? I'd start by running these sort of tests using phantomjs which has the ghostdriver built-in. a) Add the following packages: sudo pkg_add jre sudo pkg_add phantomjs sudo pkg_add python-3.3.2 b) Fetch Selenium and install the Python module sudo pip-3.3 install selenium ftp http://selenium-release.storage.googleapis.com/2.42/selenium-server-standalone-2.42.2.jar c) Fire up the server java -jar selenium-server-standalone-2.42.2.jar phantomjs --webdriver=8080 --webdriver-selenium-grid-hub=http://127.0.0.1: d) Write a test! #!/usr/local/bin/python3.3 from selenium import webdriver browser = webdriver.PhantomJS(executable_path=/usr/local/bin/phantomjs) browser.get(http://openbsd.org/;) assert OpenBSD in browser.title print(DONE) -- Eric Radman
Re: new: pop3d-1.0
On Sat, May 31, 2014 at 12:48:55PM -0400, James Turner wrote: Attached is a new port for pop3d. Tested on amd64. Also tested on macppc, and using in production on i386 You might remember this as the daemon Sunil sent to tech@ awhile back. I've been running it in production with great success. It was developed specifically for OpenBSD in the same vein as smtpd. It will require a new _pop3d user/group which would be 733 at the time of this mailing. For POP3S you need a server.crt and server.key file. The man page covers this so I didn't feel a README file was necessary, but I can add one if needed. oks? To ensure that setting CFLAGS in mk.conf doesn't break this port I think the preprocessor directives should be set with CPPFLAGS: --- Makefile.orig Sat May 31 22:35:55 2014 +++ MakefileSat May 31 23:12:32 2014 @@ -3,7 +3,7 @@ MAN=pop3d.8 CFLAGS+= -Wall -Wstrict-prototypes -Wmissing-prototypes CFLAGS+= -Wmissing-declarations -Wshadow -Wpointer-arith CFLAGS+= -Wcast-qual -Wsign-compare -CFLAGS+= -DIO_SSL +CPPFLAGS+= -DIO_SSL DEBUG= -g SRCS= pop3d.c pop3e.c session.c maildrop.c maildir.c mbox.c util.c SRCS+= imsgev.c iobuf.c ioev.c
akpop3d does not work out of the box
The upgrade notes for 5.5 suggest that akpop3d might work as a
replacement for the popa3d that was removed from base. Is there a trick
to using this daemon? akpop3d does not seem to manage locks correctly:
# tail /var/log/maillog
May 23 23:27:16 vm akpop3d[4954]: Connection from 127.0.0.1:27515
May 23 23:27:16 vm akpop3d[6121]: Authenticated eradman
May 23 23:27:21 vm akpop3d[6121]: failed to lock maildrop:
/var/mail/eradman:
File exists
# ls /var/mail/eradman*
-rw--- 1 eradman users 2334 May 3 23:28 eradman
-rw-r- 1 eradman _akpop3d 5 May 26 08:38 eradman.lock
I ended up using solid-pop3d which was trivial to install. I've been
using nginx from ports to provide SSL access:
# pf.conf
block in on ! lo0 proto tcp to port pop3
# nginx.conf
mail {
server_name vm.eradman.com;
auth_http localhost:9000;
proxy on;
ssl_protocols TLSv1 SSLv3;
ssl_certificate /etc/mail/certs/vm.eradman.com.crt;
ssl_certificate_key /etc/mail/certs/vm.eradman.com.key;
pop3_auth plain apop cram-md5;
server {
protocolpop3;
listen 995;
ssl on;
pop3_auth plain;
}
}
--
Eric Radman
Re: NEW: textproc/igor
On Thu, May 15, 2014 at 03:40:37AM +0200, Ingo Schwarze wrote: Hi, i've just met Warren Block at the FreeBSD DocSprint during BSDCan and he has shown me his igor utility. This may be useful to complement mandoc -Tlint and mdoclint to check manuals for markup errors, and especially for typos and stilistic issues. OK? Ingo This is useful. I tried this on several files, and it caught some real mistakes I had made in xhtml and mdoc. Eric
NEW: graphics/pngquant
Attached is a port for PNG compressor that uses a lossy algorithm that I've found to be very effective. The maintainer of this project mode a few modifications to make this project easier to build on OpenBSD (https://github.com/pornel/pngquant/issues/52) Tested on i386 and ppc. -- Eric Radman | http://eradman.com pngquant_port.tar.gz Description: application/tar-gz
Re: NEW: graphics/pngquant
On Wed, Apr 23, 2014 at 10:44:25AM -0400, Brian Callahan wrote:
On 04/23/14 10:19, Eric Radman wrote:
Attached is a port for PNG compressor that uses a lossy algorithm that
I've found to be very effective.
Couple of notes:
You'll need MAKE_FLAGS=CC=${CC} since the port sets CC to gcc.
I'd rather see you use a do-install for the binary and man page (or make
your patch do the right thing) since neither are installed with the
proper group or mode.
Good, added MAKE_FLAGS= and confirmed it's respecting the compiler set
in mk.conf. Added do-install and removed my previous changes to the
install section of the Makefile.
I'm not 100% sure your patch-Makefile is correct. As it is, this
installs the debug version of pngquant: the CFLAGS?= line gets
overridden with the default CFLAGS. Did you mean to also make that
CFLAGS+= like the line below it? That will make it build the
non-debug pngquant. (Don't forget to remove the -O3 from the CFLAGSOPT
line as well). CUSTOMLIBPNG and CUSTOMZLIB can go away as well.
The updated port (attached) uses CFLAGS+= with the DEBUG and -03 options
removed. Also removed the lines that set CUSTOMLIBPNG and CUSTOMZLIB.
Thanks for the feeback!
Eric
pngquant_port.tar.gz
Description: application/tar-gz
UPDATE: sysutils/entr 1.9 - 2.8
Greetings,
This is an update to sysutils/entr. Notable changes since 1.9:
* Wait for processes to terminate in restart mode
* Prevent interactive utilities from paging output by setting PAGER to
`/bin/cat`
* First match of `/_` provides a shortcut for specifying the first file
that changed
* New `-c` option invokes `clear` before running the utility
* Support Vim's `backup` option by responding to `NOTE_RENAME`
* Iteratively check for new file system events using a 1ms timeout after
a file has been renamed or removed in order to prevent the utility
from being executed before successive changes from a version control
system have completed
Index: ports/sysutils/entr/Makefile
===
RCS file: /cvs/ports/sysutils/entr/Makefile,v
retrieving revision 1.5
diff -u -p -r1.5 Makefile
--- ports/sysutils/entr/Makefile23 Apr 2013 23:04:45 - 1.5
+++ ports/sysutils/entr/Makefile15 Mar 2014 18:06:19 -
@@ -1,9 +1,9 @@
-# $OpenBSD: Makefile,v 1.5 2013/04/23 23:04:45 jturner Exp $
+# $OpenBSD: Makefile,v $
COMMENT = run arbitrary commands when files change
-DISTNAME = entr-1.9
-REV = 587c18272485
+DISTNAME = entr-2.8
+REV = 9192ece7198c
CATEGORIES = sysutils
@@ -22,6 +22,7 @@ MASTER_SITES = http://entrproject.org/co
MAKE_FILE = Makefile.bsd
MAKE_FLAGS = CC=${CC} LDFLAGS=${LDFLAGS}
WRKDIST = ${WRKDIR}/eradman-entr-${REV}
+FAKE_FLAGS = DESTDIR=
TEST_TARGET = test
Index: ports/sysutils/entr/distinfo
===
RCS file: /cvs/ports/sysutils/entr/distinfo,v
retrieving revision 1.3
diff -u -p -r1.3 distinfo
--- ports/sysutils/entr/distinfo23 Apr 2013 23:04:45 - 1.3
+++ ports/sysutils/entr/distinfo15 Mar 2014 18:06:19 -
@@ -1,2 +1,2 @@
-SHA256 (entr-1.9.tar.gz) = ek3XfDaUrIihDQ0eRoyn8SaCdcQZBD2dNVFO9nY+c5Y=
-SIZE (entr-1.9.tar.gz) = 8680
+SHA256 (entr-2.8.tar.gz) = hHq+LfbIV68OA+P259Raact3qDoz52gIKUn2cWH9RMw=
+SIZE (entr-2.8.tar.gz) = 20185
Re: firefox libstdc++ and libestdc++ problem
On Mon, Mar 03, 2014 at 11:38:40AM -0600, Chris Bennett wrote: I have had this problem for a while, but not sure how to fix. Latest snap. ... see following: firefox:/usr/lib/libstdc++.so.57.0: /usr/local/lib/libestdc++.so.14.0 : WARNING: symbol(_ZN11__gnu_debug17_S_debug_messagesE) size mismatch, relink your program ed 5.5current # ls /usr/lib/libstdc++* -r--r--r-- 1 root bin 8772138 Mar 3 01:46 /usr/lib/libstdc++.a -r--r--r-- 1 root bin 4889709 Jul 28 2013 /usr/lib/libstdc++.so.55.0 -r--r--r-- 1 root bin 4888944 Mar 3 01:46 /usr/lib/libstdc++.so.57.0 -r--r--r-- 1 root bin 8533454 Mar 3 01:46 /usr/lib/libstdc++_p.a ed 5.5current # ls /usr/local/lib/libestdc++.so.14.0 -rwxr-xr-x 1 root bin 4871462 Jan 13 04:58 /usr/local/lib/libestdc++.so.14.0 What should I do to resolve this? This was raised almost a year ago, see http://marc.info/?l=openbsd-portsm=136490780102023w=2 As far as I know it's a superfluous warning imposed by the GNU toolchain
Re: Patch firefox-25.0 to enable gssapi/Kerberos
On Sat, Jan 18, 2014 at 09:59:13PM +0100, Landry Breuil wrote:
On Fri, Jan 17, 2014 at 11:21:13AM -0500, Eric Radman wrote:
Resubmitting a patch for firefox that includes all of the symbols
required to load gssapi. From moz.log:
784166112[83e29180]: service = inout.deshaw.com
784166112[83e29180]: using negotiate-gss
784166112[83e29180]: entering nsAuthGSSAPI::nsAuthGSSAPI()
784166112[83e29180]: Attempting to load gss functions
784166112[83e29180]: entering nsAuthGSSAPI::Init()
784166112[83e29180]: nsHttpNegotiateAuth::GenerateCredentials()
[challenge=Negotiate]
784166112[83e29180]: entering nsAuthGSSAPI::GetNextToken()
784166112[83e29180]: leaving nsAuthGSSAPI::GetNextToken [rv=0]
784166112[83e29180]: Sending a token of length 1730
Notes on the patch:
- Remove ` !lib` from the module load loop because it would stop
loading modules the first time dlopen succeeded
- Had to replace PR_LoadLibrary with PR_LoadLibraryWithFlags so I
could use the PR_LD_GLOBAL option
So, after having had some time to actually look at your patch - i
suppose that we need to dlopen all the libs one by one in the right
order, because they're not inter-dependent in our case?
libgssapi.so doesn't appear to reference any of it's dependencies. On
many other platforms they are; this is Debian 7.0:
$ readelf -d /usr/lib/i386-linux-gnu/libgssapi_krb5.so.2 | grep NEEDED
0x0001 (NEEDED) Shared library: [libkrb5.so.3]
0x0001 (NEEDED) Shared library: [libk5crypto.so.3]
0x0001 (NEEDED) Shared library: [libcom_err.so.2]
0x0001 (NEEDED) Shared library: [libkrb5support.so.0]
0x0001 (NEEDED) Shared library: [libdl.so.2]
0x0001 (NEEDED) Shared library: [libkeyutils.so.1]
0x0001 (NEEDED) Shared library: [libresolv.so.2]
0x0001 (NEEDED) Shared library: [libc.so.6]
Why do you need RTLD_GLOBAL ? It fails if you use the default symbol
visibility flag ?
RTLD_GLOBAL seems to be required, although I don't understand why. This
is how I'm testing:
#include dlfcn.h
main()
{
dlopen(libasn1.so, RTLD_GLOBAL);
dlopen(libcrypto.so, RTLD_GLOBAL);
dlopen(libroken.so, RTLD_GLOBAL);
dlopen(libheimbase.so, RTLD_GLOBAL);
dlopen(libcom_err.so, RTLD_GLOBAL);
dlopen(libkrb5.so, RTLD_GLOBAL);
dlopen(libgssapi.so, RTLD_GLOBAL|RTLD_NOW);
}
Instead of patching Firefox, would it be better to change the way
OpenBSD links the Heimdal libraries to make the interdependent libraries
implicit?
Eric
Patch firefox-25.0 to enable gssapi/Kerberos
Resubmitting a patch for firefox that includes all of the symbols
required to load gssapi. From moz.log:
784166112[83e29180]: service = inout.deshaw.com
784166112[83e29180]: using negotiate-gss
784166112[83e29180]: entering nsAuthGSSAPI::nsAuthGSSAPI()
784166112[83e29180]: Attempting to load gss functions
784166112[83e29180]: entering nsAuthGSSAPI::Init()
784166112[83e29180]: nsHttpNegotiateAuth::GenerateCredentials()
[challenge=Negotiate]
784166112[83e29180]: entering nsAuthGSSAPI::GetNextToken()
784166112[83e29180]: leaving nsAuthGSSAPI::GetNextToken [rv=0]
784166112[83e29180]: Sending a token of length 1730
Notes on the patch:
- Remove ` !lib` from the module load loop because it would stop
loading modules the first time dlopen succeeded
- Had to replace PR_LoadLibrary with PR_LoadLibraryWithFlags so I
could use the PR_LD_GLOBAL option
$OpenBSD$
--- extensions/auth/nsAuthGSSAPI.cpp.orig Fri Oct 25 18:27:17 2013
+++ extensions/auth/nsAuthGSSAPI.cppThu Nov 21 09:14:15 2013
@@ -120,22 +120,32 @@ gssInit()
PR_FreeLibraryName(libName);
}
#else
-
+
const char *const libNames[] = {
-gss,
-gssapi_krb5,
+asn1,
+crypto,
+roken,
+heimbase,
+com_err,
+krb5,
gssapi
};
-
+
const char *const verLibNames[] = {
-libgssapi_krb5.so.2, /* MIT - FC, Suse10, Debian */
-libgssapi.so.4, /* Heimdal - Suse10, MDK */
-libgssapi.so.1, /* Heimdal - Suse9, CITI - FC, MDK, Suse10*/
-libgssapi.so /* OpenBSD */
+libasn1.so,
+libcrypto.so,
+libroken.so,
+libheimbase.so,
+libcom_err.so,
+libkrb5.so,
+libgssapi.so
};
-for (size_t i = 0; i ArrayLength(verLibNames) !lib; ++i) {
-lib = PR_LoadLibrary(verLibNames[i]);
+PRLibSpec libSpec;
+for (size_t i = 0; i ArrayLength(verLibNames); ++i) {
+libSpec.type = PR_LibSpec_Pathname;
+libSpec.value.pathname = verLibNames[i];
+lib = PR_LoadLibraryWithFlags(libSpec, PR_LD_GLOBAL);
/* The CITI libgssapi library calls exit() during
* initialization if it's not correctly configured. Try to
Re: [update] py-simplejson
On Sun, Dec 29, 2013 at 12:18:28PM +0100, frantisek holop wrote: hmm, on Sat, Dec 28, 2013 at 11:54:28PM -0800, Martynas Venckus said that The diff looks good but I don't use this anymore, so could you remove me as maintainer and find someone to take care of testing/committing it? ok, i take over. attached the new version. please test and commit. -f -- i'm here to question all your answers. Index: Makefile === RCS file: /cvs/ports/devel/py-simplejson/Makefile,v retrieving revision 1.19 diff -u -p -r1.19 Makefile --- Makefile 11 Mar 2013 10:50:26 - 1.19 +++ Makefile 29 Dec 2013 11:16:55 - @@ -2,15 +2,14 @@ COMMENT =JSON encoder/decoder -MODPY_EGG_VERSION = 2.5.0 +MODPY_EGG_VERSION = 3.3.1 simplejson is listed as a dependency for several ports, but are they using it? It has been bundled with Python since 2.6. Projects that still support Python 2.5 will typically do include some logic that tries to import the standard library first try: import json except ImportError: import simplejson as json
Using Kerberos auth with Firefox on 5.4, 5.4-current
Is there any special configuration for enabling Kerberos authentication in Firefox? Normally one would simply set the Kerberos whitelist for using about:config: network.negotiate-auth.trusted-uris = my.domain, My impression is that the implementation in Firefox is broken. Here is what I tried: 1. Enable firefox debug $ export NSPR_LOG_MODULES=negotiateauth:5 $ export NSPR_LOG_FILE=/tmp/moz.log $ ktrace firefox 2. Log shows that gssapi failed to load $ cat /tmp/moz.log 750271712[89789780]: service = inout.deshaw.com 750271712[89789780]: using negotiate-gss 750271712[89789780]: entering nsAuthGSSAPI::nsAuthGSSAPI() 750271712[89789780]: Fail to load gssapi library 750271712[89789780]: entering nsAuthGSSAPI::Init() 3. ktrace confirms the attempt to load libgssapi.so (5.4-release) $ kdump | grep -B1 -A1 gss ... 20139 firefox CALL open(0x214383e0,0O_RDONLY) 20139 firefox NAMI /usr/lib/libgssapi.so.6.0 20139 firefox RET open 51/0x33 -- 20139 firefox GIO fd 2 wrote 74 bytes firefox:/usr/lib/libgssapi.so.6.0: undefined symbol 'krb5_cc_type_memory' Using firefox-25.0 from 5.4-current (Nov 13 snapshot) I see this: 30981 firefox CALL open(0x229e43e0,0O_RDONLY) 30981 firefox NAMI /usr/lib/libgssapi.so.7.1 30981 firefox RET open 53/0x35 -- 30981 firefox GIO fd 2 wrote 74 bytes firefox:/usr/lib/libgssapi.so.7.1: undefined symbol 'krb5_cc_type_memory' 4. This symbol is definitely visible $ nm /usr/lib/libgssapi.so.7.1 | grep krb5_cc_type_memory U krb5_cc_type_memory Even without this functionality it's wonderful to see firefox-25.0 in -current! -- Eric Radman | http://eradman.com
MASTER_SITE update to wpi-firmware
The firmware for the Intel PRO/Wireless 3945ABG is no longer available
for download at intellinuxwireless.org. I couldn't find a note
explaining why they changed the download location, but it appears to be
related to an anti-hotlinking campaign.
Index: ports/sysutils/firmware/wpi/Makefile
===
RCS file: /cvs/ports/sysutils/firmware/wpi/Makefile,v
retrieving revision 1.3
diff -u -p -r1.3 Makefile
--- ports/sysutils/firmware/wpi/Makefile12 Jun 2011 00:20:51 -
1.3
+++ ports/sysutils/firmware/wpi/Makefile17 Oct 2013 14:22:04 -
@@ -3,13 +3,13 @@
FW_DRIVER= wpi
FW_VER=3.2
DISTNAME= iwlwifi-3945-ucode-15.32.2.9
-REVISION= 0
+REVISION= 1
WRKDIST= ${WRKDIR}
EXTRACT_SUFX= .tgz
-HOMEPAGE= http://www.intellinuxwireless.org/
-MASTER_SITES= ${HOMEPAGE}iwlwifi/downloads/ \
+HOMEPAGE= http://wireless.kernel.org/
+MASTER_SITES=
${HOMEPAGE}en/users/Drivers/iwlegacy?action=AttachFile\do=get\target= \
http://firmware.openbsd.org/firmware-dist/
do-install:
Re: Early Look: PostgreSQL 9.3 beta 1
On Thu, Jun 27, 2013 at 03:08:55PM -0700, Jeremy Evans wrote: Here's a diff against -current for 9.3 beta2. I've made some PLIST changes so that libs for contrib modules are in the contrib package instead of the server package, and fixed an issue in the 9.3beta1 diff where a new header file was added to the server package instead of the client package. Very nice! The major extensions just work out of the box # SELECT * FROM pg_extension; extname | extowner | extnamespace | extrelocatable | extversion | extconfig | extcondition ---+--+--+++---+-- plpgsql | 10 | 11 | f | 1.0| | plpythonu |16432 | 11 | f | 1.0| | hstore|16432 | 2200 | t | 1.1| | uuid-ossp | Since 9.3 doesn't depend on SysV shared memory, we can simplify README-server a bit, and maybe suggest a command for changing the login class --- README-server Fri Jun 28 19:42:44 2013 +++ README-server.new Fri Jun 28 19:42:53 2013 @@ -67,22 +67,10 @@ Tuning for busy servers === -The default sizes in the GENERIC kernel for SysV semaphores are only -just large enough for a database with the default configuration -(max_connections 40) if no other running processes use semaphores. -In other cases you will need to increase the limits. Adding the -following in /etc/sysctl.conf will be reasonable for many systems: - kern.seminfo.semmni=60 - kern.seminfo.semmns=1024 +Adjust the value of max_connections in postgresql.conf to support more +than 40 connections. -To serve a large number of connections (250), you may need higher -values for the above. - -You may also want to tune the max_connect value in the -postgresql.conf file to increase the number of connections to the -backend. - By default, the _postgresql user, and so the postmaster and backend processes run in the login(1) class of daemon. On a busy server, it may be advisable to put the _postgresql user and processes in @@ -98,6 +86,10 @@ Rebuild the login.conf.db file if necessary: # [ -f /etc/login.conf.db ] cap_mkdb /etc/login.conf + +Then change the login class using + + # usermod -L postgresql _postgresql For more than about 250 connections, these numbers should be increased. Please report any changes and experiences to the package
Re: Early Look: PostgreSQL 9.3 beta 1
On Fri, Jun 28, 2013 at 05:25:57PM -0700, Jeremy Evans wrote: On 06/28 07:49, Eric Radman wrote: Since 9.3 doesn't depend on SysV shared memory, we can simplify README-server a bit Are you sure the changes to SysV shared memory affect the SysV semaphore situation? I already modified the README for the SysV shared memory differences. Comparing the following two pages: http://www.postgresql.org/docs/9.2/static/kernel-resources.html http://www.postgresql.org/docs/9.3/static/kernel-resources.html It looks like the semaphore situation hasn't changed. Sorry, you are correct. I hadn't read the release notes carefully, and if I crank up max_connections the error log is informative FATAL: could not create semaphores: No space left on device DETAIL: Failed system call was semget(5432004, 17, 03600). HINT: This error does *not* mean that you have run out of disk space. It occurs when either the system limit for the maximum number of semaphore sets (SEMMNI), or the system wide maximum number of semaphores (SEMMNS), would be exceeded. You need to raise the respective kernel parameter. Alternatively, reduce PostgreSQL's consumption of semaphores by reducing its max_connections parameter.
Is ports/books useful?
Some 10 years ago the 'books' category was added to ports. I presume this category was experimental. Has it proved to be useful? If not, should it be deprecated? Consider how I might read Dive Into Python: $ cd /usr/ports/books/diveintopython/ $ sudo make install $ pkg_info -f diveintopython-5.4p0 | grep pdf $ xpdf /usr/local/share/doc/diveintopython/print/diveintopython.pdf There are some benefits to insisting that users keep such material in their home directories instead: - root privileges not required - files will still be there after the next upgrade - less space on OpenBSD mirrors - easier to locate material I've downloaded personally -- Eric Radman
Re: Is ports/books useful?
On Fri, May 31, 2013 at 02:44:16PM +0100, Stuart Henderson wrote: On 2013/05/31 08:43, Eric Radman wrote: Some 10 years ago the 'books' category was added to ports. I presume this category was experimental. Has it proved to be useful? should it be deprecated? Consider how I might read Dive Into Python: If you already know about the book, then yes of course there are other ways to read it which might be easier. But another way to think of it is as a guide to books which porters have found useful/interesting enough that they want to share them with others. Okay, from that vantage point they're useful, but in that case isn't this directory wildly incomplete? Now if there were 200 titles it could be a compelling resource. There are some benefits to insisting that users keep such material in their home directories instead: - root privileges not required - files will still be there after the next upgrade I don't understand this about upgrades, do you mean that you want to specifically keep an older version when ports has updated to a newer one? Upgrading of course doesn't remove existing packages.. Yes, `pkg_add -ui` preserves packages. But I tend to remove all packages and then reisntall the packages I need on each machine. - easier to locate material I've downloaded personally It might be useful to consolidate the installed files into a single directory (/usr/local/share/books?) I think that's a great idea.
Re: simh-3.9.0p1 on 5.3
On Thu, May 30, 2013 at 02:57:16PM +0200, Eivind Evensen wrote: On Sun, May 19, 2013 at 01:47:46AM -0400, Eric Radman wrote: On Sun, May 19, 2013 at 02:03:13AM +0100, Federico Schwindt wrote: On Sat, May 18, 2013 at 4:49 PM, Eric Radman [1]ericsh...@eradman.com wrote: Following the instructions at [2]http://www.openbsd.org/vax-simh.html ... I'm not able to establish communication using bridge(4) or bpf(4) when running simh-vax. is link0 set on tun0 as specified on the vax-simh doc? f.- Indeed. I tried it on two different systems (both i386), one using a bridge with an em(4) NIC and another with a bridge with a dc(4) NIC. Eric Is tun0 up? Yes, the instructions are pretty clear. No luck yet. $ ifconfig tun0 tun0: flags=9942BROADCAST,RUNNING,PROMISC,SIMPLEX,LINK0,MULTICAST mtu 1500 lladdr fe:e1:ba:d1:39:fa priority: 0 groups: tun status: active $ ifconfig bridge0 bridge0: flags=41UP,RUNNING groups: bridge priority 32768 hellotime 2 fwddelay 4 maxage 20 holdcnt 6 proto rstp designated: id 00:00:00:00:00:00 priority 0 em0 flags=3LEARNING,DISCOVER port 1 ifpriority 0 ifcost 0 tun0 flags=3LEARNING,DISCOVER port 6 ifpriority 0 ifcost 0 Addresses (max cache: 100, timeout: 240): ...
Re: simh-3.9.0p1 on 5.3
On Thu, May 30, 2013 at 03:43:36PM +0200, Eivind Evensen wrote: On Thu, May 30, 2013 at 09:25:10AM -0400, Eric Radman wrote: On Thu, May 30, 2013 at 02:57:16PM +0200, Eivind Evensen wrote: Is tun0 up? Yes, the instructions are pretty clear. No luck yet. $ ifconfig tun0 tun0: flags=9942BROADCAST,RUNNING,PROMISC,SIMPLEX,LINK0,MULTICAST mtu 1500 lladdr fe:e1:ba:d1:39:fa priority: 0 groups: tun status: active That doesn't look like up to me. Compare with this: .ifconfig tun90 tun90: flags=9943UP,BROADCAST,RUNNING,PROMISC,SIMPLEX,LINK0,MULTICAST mtu 1500 -- Eivind # ifconfig tun0 up That's it! Thank you. I was confusing 'status: active' with 'flags=UP -- Eric
Re: update: entr-1.9
On Fri, Apr 19, 2013 at 05:04:47PM -0400, James Turner wrote: This builds off an update Eric send to ports@ for 1.8 back in January. Patch was removed so apply with -E. Tested on amd64. Hopefully Eric will take a look then I'll be looking for oks? That's almost exactly what I had in the openbsd-wip git repo. Patch applies and works fine for me on i386 and macppc. Eric Radman 1.9 New auto-reload option 2013-04-13 1.8 Loosing a file under watch is always fatal 2012-12-05 -- James Turner [-- Attachment #2: entr-1.9.diff --] [-- Type: text/plain, Encoding: 7bit, Size: 2.9K --] Index: Makefile === ...
Re: NEW graphics/rawtherapee
On Thu, Jan 24, 2013 at 08:04:02PM +0300, Kirill Bychkov wrote: Hi. Attached is a port for rawtherepee, RAW image processing application. awTherapee is a powerful RAW image processing appliction. It helps you to get the most details and least artifacts from your raw photos. Advanced color handling, enhanced exposure and tonality tools, multiple denoising methods gives you perfect results. Works fine on amd64 with Pentax RAW photos (*.pex). Nice. Also seems to work fine on i386. -- Eric Radman | http://eradman.com
Re: [DIFF] security/vpnc - annoying 'which: ip: Command not found.'
On Sun, Jan 13, 2013 at 10:05:05AM +, Stuart Henderson wrote: On 2013/01/12 15:00, Jiri B wrote: Just removing annoying linuxism... Please try the vpnc-script from the openconnect package (installs as /etc/openconnect/vpnc-script). If it works OK with vpnc (which it should) then I would prefer to just sync the two. # cp /etc/openconnect/vpnc-script /etc/vpnc/ Works well for me -- E. Radman
[update] sysutils/entr 1.4 - 1.8
Hi -
Here is an update to entr(1). Changes include:
- No longer use pthreads.
- Support interactive applications by opening a TTY
- Stat files that dissapear before running a command
- Always exit after failing to reopen a file under watch
E. Radman
Index: entr/Makefile
===
RCS file: /cvs/ports/sysutils/entr/Makefile,v
retrieving revision 1.2
diff -u -p -d -u entr/Makefile
--- entr/Makefile 25 May 2012 12:46:40 - 1.2
+++ entr/Makefile 11 Jan 2013 16:37:34 -
@@ -1,9 +1,9 @@
-# $OpenBSD: Makefile,v 1.2 2012/05/25 12:46:40 jasper Exp $
+# $OpenBSD: $
COMMENT = run arbitrary commands when files change
-DISTNAME = entr-1.4
-REV = b90825a157a4
+DISTNAME = entr-1.8
+REV = f508b5eb0260
CATEGORIES = sysutils
@@ -17,12 +17,12 @@ PERMIT_PACKAGE_FTP = Yes
PERMIT_DISTFILES_CDROM = Yes
PERMIT_DISTFILES_FTP = Yes
-WANTLIB += c pthread
+WANTLIB += c
-MASTER_SITES = http://static.eradman.com/code/ \
+MASTER_SITES = http://entrproject.org/code/ \
https://bitbucket.org/eradman/entr/get/
-MAKE_FLAGS = CC=${CC} LDFLAGS=${LDFLAGS} -pthread
+MAKE_FLAGS = CC=${CC} LDFLAGS=${LDFLAGS}
WRKDIST = ${WRKDIR}/eradman-entr-${REV}
REGRESS_TARGET = test
Index: entr/distinfo
===
RCS file: /cvs/ports/sysutils/entr/distinfo,v
retrieving revision 1.2
diff -u -p -d -u entr/distinfo
--- entr/distinfo 25 May 2012 12:46:40 - 1.2
+++ entr/distinfo 11 Jan 2013 16:37:34 -
@@ -1,5 +1,2 @@
-MD5 (entr-1.4.tar.gz) = gDUdLW+XuWS6EkIScz3nnQ==
-RMD160 (entr-1.4.tar.gz) = Ky5g5HeoMa714W/Pg7AUFIdP9Rk=
-SHA1 (entr-1.4.tar.gz) = KBZODf88PIZ/XjC0rQ3oOd+Mjkc=
-SHA256 (entr-1.4.tar.gz) = /ARp/wYGI026FGg9YKoWM/xgz/p8nQQU/ALdmjv+SoE=
-SIZE (entr-1.4.tar.gz) = 7240
+SHA256 (entr-1.8.tar.gz) = COdYzAVlbhQuvN5cK0nhAgX9Fuy6zuQgvGkr1DGm+Hw=
+SIZE (entr-1.8.tar.gz) = 7936
Index: entr/pkg/DESCR
===
RCS file: /cvs/ports/sysutils/entr/pkg/DESCR,v
retrieving revision 1.1.1.1
diff -u -p -d -u entr/pkg/DESCR
--- entr/pkg/DESCR 12 Apr 2012 14:55:47 - 1.1.1.1
+++ entr/pkg/DESCR 11 Jan 2013 16:37:34 -
@@ -1,3 +1,3 @@
entr - a utility for running arbitrary commands when files change.
-Uses kqueue(2) to avoid polling. Reads a list of files provided on
-STDIN and runs the supplied command if any of them are modified.
+Uses kqueue(2) to avoid polling. entr responds to file system events
+by executing command line arguments or by writing to a FIFO.
Re: UPDATE: x11/st
On Sun, Nov 04, 2012 at 05:53:08PM +0100, Joerg Zinke wrote: Hi, please find below an update for x11/st to version 0.3. From the suckless announcement, the biggest changes are: * New drawing code, which is way more faster and comparable to the other terminals out there. * Many new escape sequences are supported: * setting the title * LNM, KAM, SI, SO, SUB, CAN, DA, DECID. OSC, DSC, PM, APC, CHT, HTS, CBT, SM, RM, standout mode ??? (See st.info.) * Linedrawing is now supported * Xft and Fontcache is now used to draw fonts. * support for italic and bold italic font drawing * Some more commandline parameters are available. (See manpage.) * easy debugging with: st -o - | cat -v * nyancat(1) works. * external support of utmp entries by using utmp(1)[1] Comments and OKs? They've made significant progress, nice to see. Minor comments: * It would be nice of the man page indicated how how to change the font name AND size: st -f Anonymous Pro:pixelsize=20 * Perhaps remove -Os from the CFLAGS More significantly, characters 127 aren't rendered. st doesn't appear to have this problem on Linux. Eric
Re: NEW: sysutils/entr
On Thu, Apr 12, 2012 at 02:50:42PM +0100, Stuart Henderson wrote: On 2012/04/12 08:38, Eric Radman wrote: Event Notify Test Runner entr is a utility for running arbitrary commands when files change. Uses kqueue(2) to avoid polling. Reads a list of files provided on STDIN and runs the supplied command if any of them are modified. See https://bitbucket.org/eradman/entr/overview - missing # $OpenBSD$ line - use REGRESS_TARGET=test rather than custom do-regress - can you run DESCR through fmt please, layout is much nicer that way Thanks; advice taken, new port attached. I also re-tagged entr-1.0 in the repository with a Makefile that respects CFLAGS. -- Eric Radman | http://eradman.com entr-1.0.tgz Description: application/tar-gz
Regression tests for PCC
I would like to include regression tests for the new port of PCC
https://github.com/jasperla/openbsd-wip/tree/master/lang/pcc
What is the correct way to include external software for regression
tests? I know this isn't right...
do-regress:
# ftp -C http://example.org/pcc-tests.tgz
# tar zxvf pcc-tests.tgz
# CC=${WRKBUILD}/cc/cc/pcc cd pcc-tests ./build-and-run-tests.sh
Eric
Re: src/usr.bin/pcc/
On Thu, Dec 22, 2011 at 09:00:34PM -0600, Amit Kulkarni wrote: On Thu, Dec 22, 2011 at 2:28 PM, Eric Radman ericsh...@eradman.com wrote: I noticed that src/usr.bin/pcc/ is quite old, (amd64 is missing). Are there any plans to update the source tree, or should we make a new port? lang/pcc usr.bin/pcc will get updated eventually. you are welcome to submit a port and see if there's any interest in getting a 1.0 port committed. Very good, I'll submit a new port for pcc 1.0.
src/usr.bin/pcc/
I noticed that src/usr.bin/pcc/ is quite old, (amd64 is missing). Are there any plans to update the source tree, or should we make a new port? lang/pcc Eric Radman | http://eradman.com
