Re: MAKEFLAGS caused temporary insanity

2015-12-05 Thread Jan Beich
Bryan Drewery  writes:

> On 11/24/2015 8:00 PM, Euan Thoms wrote:
>
>>  
>> On Tuesday, November 24, 2015 11:15 SGT, Jan Beich  
>> wrote: 
>>  
>>> "Euan Thoms"  writes:
>>>
 do-build:
 cd ${WRKSRC} && unset MAKEFLAGS && ${OPENSIPS_MAKE_ENV}
 ${GMAKE} ${_MAKE_JOBS} ${ALL_TARGET}
>>>
>>> Have you tried defining NOPRECIOUSMAKEVARS in the port's Makefile?
>>>
>> 
>> That did the trick, thanks. I can now remove the "unset MAKEFLAGS" in 
>> do-build.
>>  
>>  
>
> As far as I can tell NOPRECIOUSMAKEVARS does not impact the variable
> literally named 'MAKEFLAGS'. It should only be impacting these:
>
> # make -V _EXPORTED_VARS
> ARCH OPSYS _OSRELEASE OSREL OSVERSION _PKG_CHECKED HAVE_COMPAT_IA32_KERN
> _SMP_CPUS CONFIGURE_MAX_CMD_LEN

Did you read my reply untrimmed? net/opensips uses ARCH in vendor Makefile
and the framework clobbers its value. Some ports like multimedia/ffmpeg
rename the variable, others give up and define NOPRECIOUSMAKEVARS.

> Which port was this? Which release are you running (real question is
> whether it was fmake or bmake)?

Affects both. I did propose 2 ways to fix.


signature.asc
Description: PGP signature


Fwd: Memory leak or misconfiguration: Repeated allocation of very large block

2015-12-05 Thread Sebastian Wolfgarten
Any views on this?

> Anfang der weitergeleiteten Nachricht:
> 
> Von: Sebastian Wolfgarten 
> Datum: 25. Oktober 2015 um 11:55:42 MEZ
> An: po...@freebsd.org
> Betreff: Wtr: Memory leak or misconfiguration: Repeated allocation of very 
> large block
> 
> To whom it may concern,
> 
> in view of the discussion below may I suggest we remove the --with-gc option 
> from the Dovecot default options?
> 
> Thanks.
> 
> Kind regards
> Sebastian
> 
>> Anfang der weitergeleiteten Nachricht:
>> 
>> Von: Sebastian Wolfgarten > >
>> Datum: 25. Oktober 2015 um 11:54:15 MEZ
>> An: Timo Sirainen mailto:t...@iki.fi>>
>> Kopie: dove...@dovecot.org 
>> Betreff: Aw: Memory leak or misconfiguration: Repeated allocation of very 
>> large block
>> 
>> Hi Timo,
>> 
>> alright, thanks I will recompile without gc - this is a standard build 
>> option on FreeBSD but I will get the port maintainer to maybe change this.
>> 
>> Best regards
>> Sebastian
>> 
>>> Am 24.10.2015 um 23:19 schrieb Timo Sirainen >> >:
>>> 
>>> On 24 Oct 2015, at 14:59, Sebastian Wolfgarten >> > wrote:
 
 Dear all,
 
 I am using Dovecot 2.2.19 and I keep on getting the following warnings in 
 my mail.log file on FreeBSD 10:
 
 Oct 24 10:45:28 server1 dovecot: imap: Error: GC Warning: Repeated 
 allocation of very large block (appr. size 20480):
 Oct 24 10:45:28 server1 dovecot: imap: Error:  May lead to memory leak 
 and poor performance.
 
 I am wondering whether this is a configuration issue (i.e. a mistake in my 
 current config) or whether I am hitting a bug. What’s the best way of 
 figuring this one out? I tried increasing the verbosity of the logs but it 
 did not really provide me with further information.
>>> 
>>> I guess you have configured Dovecot --with-gc? I guess it's possible this 
>>> is a bug, but it could just as well be a false alarm. In general I don't 
>>> recommend/support --with-gc option (and I should probably just remove it 
>>> altogether). The easiest and the most usable way of finding memory leaks is 
>>> to run with valgrind (which must be done without-gc):
>>> 
>>> service imap {
>>> executable = /usr/bin/valgrind -q --show-reachable=yes --leak-check=full 
>>> /usr/libexec/dovecot/imap
>>> }
>>> 
>>> I'm not aware of any memory leaks right now.
>>> 
>> 
> 



signature.asc
Description: Message signed with OpenPGP using GPGMail


Re: MAKEFLAGS caused temporary insanity

2015-12-05 Thread Bryan Drewery
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512

On 12/5/15 1:09 AM, Jan Beich wrote:
> Bryan Drewery  writes:
> 
>> On 11/24/2015 8:00 PM, Euan Thoms wrote:
>> 
>>> 
>>> On Tuesday, November 24, 2015 11:15 SGT, Jan Beich
>>>  wrote:
>>> 
 "Euan Thoms"  writes:
 
> do-build: cd ${WRKSRC} && unset MAKEFLAGS &&
> ${OPENSIPS_MAKE_ENV} ${GMAKE} ${_MAKE_JOBS} ${ALL_TARGET}
 
 Have you tried defining NOPRECIOUSMAKEVARS in the port's
 Makefile?
 
>>> 
>>> That did the trick, thanks. I can now remove the "unset
>>> MAKEFLAGS" in do-build.
>>> 
>>> 
>> 
>> As far as I can tell NOPRECIOUSMAKEVARS does not impact the
>> variable literally named 'MAKEFLAGS'. It should only be impacting
>> these:
>> 
>> # make -V _EXPORTED_VARS ARCH OPSYS _OSRELEASE OSREL OSVERSION
>> _PKG_CHECKED HAVE_COMPAT_IA32_KERN _SMP_CPUS
>> CONFIGURE_MAX_CMD_LEN
> 
> Did you read my reply untrimmed? net/opensips uses ARCH in vendor
> Makefile

This is what I was looking for.

> and the framework clobbers its value. Some ports like
> multimedia/ffmpeg rename the variable, others give up and define
> NOPRECIOUSMAKEVARS.
> 
>> Which port was this? Which release are you running (real question
>> is whether it was fmake or bmake)?
> 
> Affects both. I did propose 2 ways to fix.
> 

The real fix would be to not export these to the port builds. We
experimented with using 'env -i' in the build a few years back and had
a lot of fallout.

- -- 
Regards,
Bryan Drewery
-BEGIN PGP SIGNATURE-
Comment: GPGTools - https://gpgtools.org

iQEcBAEBCgAGBQJWYxi/AAoJEDXXcbtuRpfPghcIAMOzkIFfm4+s7MXi2/B6UrqB
r9wvv9zsFZ5ytASFhuecm+47ZWkJq3sKmlw3vXHqRJRR8V9r7vhGBSINfwrOTiUE
ZMIk29/tJ1Bkdl9SDt34xU38WZD1gz+35qNZQbrpr6g1+2qGr2iX/UEOBGp6Bzrl
O6EAK18a34cbOGSwMCVxZoLhsHD2fjbLnGRROILKztE+h3s4Q2P+phQCzSGnL0vD
DqY4dLi5mKJgOj68xNJQeQKZm10+Dt+F8zdSspA6coDdnU4uRe5YxLxNLeJXpHav
CsxYLZCrYvvYaMhXHL69Lq70UdWR3L2k6bZkC/QjZlJeFNMZ6WjGSgCXrG9VWdQ=
=CJVz
-END PGP SIGNATURE-
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"


Re: Python and SWIG support in ports?

2015-12-05 Thread Craig Rodrigues
On Fri, Dec 4, 2015 at 2:40 PM, Craig Rodrigues  wrote:

> Hi,
>
> I am working with the upstream maintainer of M2Crypto (
> https://gitlab.com/m2crypto/m2crypto ).
>
> In the distutils that comes with Python, the swig binary is harcoded
> to "swig" if on a POSIX system:
>
>
> https://hg.python.org/cpython/file/v2.6.2/Lib/distutils/command/build_ext.py#l635
>
>
OK, so after reading this e-mail:
https://mail.python.org/pipermail/distutils-sig/2004-October/004204.html

I see that it is possible to override the value of "swig" by doing:

python setup.py build_ext --swig=swig2.0

I tried that for M2Crypto and it works.

Is that what the existing Python ports with SWIG are doing?

--
Craig
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"


INDEX build failed for 9.x

2015-12-05 Thread Ports Index build
INDEX build failed with errors:
Generating INDEX-9 - please wait..--- describe.accessibility ---
--- describe.arabic ---
--- describe.archivers ---
--- describe.astro ---
--- describe.audio ---
--- describe.benchmarks ---
--- describe.biology ---
--- describe.cad ---
--- describe.chinese ---
--- describe.comms ---
--- describe.converters ---
--- describe.databases ---
--- describe.deskutils ---
--- describe.devel ---
--- describe.dns ---
--- describe.editors ---
--- describe.emulators ---
--- describe.finance ---
--- describe.french ---
--- describe.ftp ---
[...]
--- describe.print ---
--- describe.russian ---
--- describe.science ---
--- describe.security ---
--- describe.shells ---
--- describe.sysutils ---
--- describe.textproc ---
--- describe.ukrainian ---
--- describe.vietnamese ---
--- describe.www ---
--- describe.x11 ---
--- describe.x11-clocks ---
--- describe.x11-drivers ---
--- describe.x11-fm ---
--- describe.x11-fonts ---
--- describe.x11-servers ---
--- describe.x11-themes ---
--- describe.x11-toolkits ---
--- describe.x11-wm ---
 Done.
make_index: 
/home/indexbuild/tindex/ports/devel/p5-Log-Dispatch-ArrayWithLimits: no entry 
for /home/indexbuild/tindex/ports/devel/p5-Perl-osnames

Committers on the hook:
 gerald nemysis pi riggs 

Most recent SVN update was:
Updating '.':
Unet/Makefile
Anet/p5-IO-Socket-Timeout
Anet/p5-IO-Socket-Timeout/Makefile
Anet/p5-IO-Socket-Timeout/distinfo
Anet/p5-IO-Socket-Timeout/pkg-descr
Anet/p5-IO-Socket-Timeout/pkg-plist
Unet/nanomsg/Makefile
Unet/nanomsg/distinfo
Unet/nanomsg/pkg-plist
Udevel/Makefile
Adevel/p5-PerlIO-via-Timeout
Adevel/p5-PerlIO-via-Timeout/Makefile
Adevel/p5-PerlIO-via-Timeout/distinfo
Adevel/p5-PerlIO-via-Timeout/pkg-descr
Adevel/p5-PerlIO-via-Timeout/pkg-plist
Ugames/lugaru/Makefile
Ugames/lugaru/pkg-descr
Ulang/gcc5/Makefile
Ulang/gcc5/distinfo
Ulang/gcc5/files/patch-arm-unwind-cxx-support
Ulang/gcc5/files/patch-unwind-ia64.h
Udatabases/cego/Makefile
Udatabases/cego/distinfo
Updated to revision 403077.
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"


INDEX now builds successfully on 9.x

2015-12-05 Thread Ports Index build

___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"


FreeBSD Port: lighttpd-1.4.37

2015-12-05 Thread Dutchman01
We need lighttpd-1.4.38 as soon as possible because of fixed bug

 

Can you please upgrade this port?

 

Regards,

Dutchman01

___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"


FreeBSD Port: php56-5.6.16

2015-12-05 Thread Dutchman01
PHP 7.0.0 has been released.

 

Can we see this soon in ports too?

 

http://php.net/downloads.php

 

Regards,

Dutchman01

___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"


pkg regression 1.6.1 to 1.6.2: custom package dependencies missing after install

2015-12-05 Thread Nick Rogers
Hello,

I have an interesting problem since upgrading to the latest pkg version. I
am running FreeBSD 10.1-RELEASE-p24.

I use "pkg create" to create a custom package (lets call it "mypkg") to
distribute some proprietary software. In the custom package's manifest, I
specify dependencies that the software relies on. These dependencies are
all "real" ports in the FreeBSD ports tree. I use poudriere to host a
custom repository that contains all the necessary dependencies. When I
bring up a new system, to install the software I run "pkg install mypkg",
and that automatically installs all the necessary dependency packages. The
custom repo is configured in /usr/local/etc/pkg/repos, and the default
"FreeBSD" repo is disabled.

This has worked great for some time until I upgraded to pkg 1.6.2
(previously was using 1.6.1). Now when I install my package for the first
time, pkg seems to ignore some of the dependencies and does not install
some packages (about 14 out of over 200).

The MANIFEST of the custom package used in pkg create looks something like
this:
Generating +MANIFEST
name: mypkg
origin: myorigin/mypkg
version: "1.23"
comment: mypkg
prefix: /mypkg
arch: freebsd:10:*
deps: {
   "bonnie++": {
  origin: "benchmarks/bonnie++"
  version: "*"
   },
   "iperf": {
  origin: "benchmarks/iperf"
  version: "*"
   },
   "mgetty": {
  origin: "comms/mgetty+sendfax"
  version: "*"
   },
   "bash": {
  origin: "shells/bash"
  version: "*"
   },
 more dependencies 


There are no errors when I do the initial pkg install mypkg. It does not
complain about missing dependencies. Most of the dependencies are
installed, and a few are left out.

When I run "pkg check -d" after the initial "pkg install mypkg", pkg
clearly recognizes that some packages are missing, but then it crashes with
an abort trap. Here you can see which packages are ignored and the error.

Checking all packages: 100%
mypkg has a missing dependency: iperf
mypkg has a missing dependency: bash
mypkg has a missing dependency: gmake
mypkg has a missing dependency: rubygem-daemons
mypkg has a missing dependency: mrtg
mypkg has a missing dependency: python
mypkg has a missing dependency: dmidecode
mypkg has a missing dependency: bwm-ng
mypkg has a missing dependency: freeradius3
mypkg has a missing dependency: flex
mypkg has a missing dependency: p5-Devel-Size
mypkg has a missing dependency: tclGetOpts

>>> Missing package dependencies were detected.
>>> Found 13 issue(s) in the package database.

Assertion failed: (pattern != NULL), function pkg_jobs_maybe_match_file,
file pkg_jobs.c, line 188.
Child process pid=29412 terminated abnormally: Abort trap

If I try the same "pkg install mypkg" with pkg 1.6.1, against the exact
same custom package file and the same poudriere repo, all packages are
installed as expected and "pkg check -d" returns a success. So it seems
that the problem lies in some change in the behavior of pkg between 1.6.1
and 1.6.2, and not my custom package or the repo. Note that the custom
package is created with pkg 1.6.2 as well.

I am inclined to think the problem has something to do with the way I am
specifying dependencies in my custom packages MANIFEST. My gut tells me it
has something to do with the following commit/change to pkg between 1.6.2
and 1.6.1, but I am unfamiliar with the code.
https://github.com/freebsd/pkg/commit/7e5152d942cba492142c7a226974913b9d38af21

The list of packages that are left-out is the same every time, but I do not
notice any correlation between them as far as how they are named or what
not, but there must be some similarity.

Hoping someone can shed some light on what is going on.

I have opened a similar issue on github in case this is the wrong forum for
such weirdness.
https://github.com/freebsd/pkg/issues/1356

Thanks!

-Nick
___
freebsd-ports@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-ports
To unsubscribe, send any mail to "freebsd-ports-unsubscr...@freebsd.org"