Hello.
I have read up on the behavior of that area for other reasons. But I've already
started to forget :)
As I recall, readelf -d or its compatible operation were used for those
searches, not ldd, so the message must mean that the wrong link was made.
So it is possible that somewhere -lfoo.n
Hello.
Probably because databases/sqlite3 defines FLAVORS but no FLAVOR is written in
*_DEPENDS of Mk/Uses/sqlite.mk like libsqlite3.so:databases/sqlite3@default.
I think python's FLAVOR is a perfect FLAVOR implementation, but sqlite3's
FLAVOR is a branch that should not use FLAVOR :)
Regards.
So is it simply because port suddenly defines FLAVOR, or is it because some
port, installed by pkg-1.17.x, is uninstalled by pkg-1.18.x?
Mk/Uses/sqlite.mk was not relevant :)
Regards.
Steve Kargl wrote on 2022/07/21 13:41:
> I'm in flavor of not breaking the ports system.
> Once again, removing
Hello.
So no matter how hard I tried I could not reproduce it, but that was a problem
that had already been fixed 😹
The portmaster must have been updated from 3.21 to 3.22 during the update that
caused the problem.
FreeBSD Bugzilla – Bug 265084 was a report of it.
Regards.
Hello.
poudriere will work more reliably if the following are written in
/usr/local/etc/poudriere.d/make.conf
OPTIONS_SET+=GSSAPI_HEIMDAL
security_ca_root_nss_UNSET+=ETCSYMLINK
To begin with, poudriere options are cumbersome :)
For example, the options created below are used for both of the fol
Hello.
query command in pkg-1.18.3 doesn't seem to properly match evaluations
containing %#r.
For example,
pkg query -e "%#r != 1" "%n-%v:%#r" | grep :1\$
At this time, it is likely that flavor is involved in all of the entries that
appear even though they should not match.
However, I am goin
Hello.
In git-2.37.1/http.c...
　　　︙
if (!curl_ssl_verify) {
curl_easy_setopt(result, CURLOPT_SSL_VERIFYPEER, 0);
curl_easy_setopt(result, CURLOPT_SSL_VERIFYHOST, 0);
} else {
　　　︙
if (getenv("GIT_SSL_NO_VERIFY"))
curl
Helge Oldach wrote on 2022/08/03 23:20:
> Maybe a port with multiple flavors just counts as a single dependency
> %#r? Just a wild guess though.
>
This is probably because they only check origin matches.
I see such SQL when I run it with pkg - query.
DBG(4)[70361]> Pkgdb:
I looked into this wondering why the poudriere is not blocking the propagation
of the value, but the poudriere is not being affected by the value.
Then, I think that is the reason why we are not getting the results we need :)
Of course, a better solution would be to set up the verification so tha
Looking at "pkg shell .dump | grep \ deps\ " results, it seems that origin
without @flavor is not unique.
However, ports that use FLAVORS are modified to make the PKGNAME unique.
So, the following patch is applied and it seems to be working properly.
The commands used in the tests are as follows.
Thank you very much for committing.
... Also, I'll shorten the subject :)
Regards.
Baptiste Daroussin wrote on 2022/08/04 20:58:
> Good catch, thanks, pushed in the repo:
> https://github.com/freebsd/pkg/commit/a1f6f620b4671d924f5fad3f24c2d7abc3b229cb
>
> Will be in the next release of pkg!!
Hello.
It seems to require cmake and ninja.
However, it doesn't seem to exist because those builds have probably stopped.
My internet connection is not set up to see the official poudriere logs, though
:)
Regards.
Yuri wrote on 2022/08/05 10:51:
> When I look at the xsimd package's Fresh ports
Yuri wrote on 2022/08/05 14:58:
> On 8/4/22 20:37, Tatsuki Makino wrote:
>> It seems to require cmake and ninja.
>
> This shouldn't affect NOARCH packages. Once such package is built on one
> architecture it should be promoted to all architectures, even to ones where
&g
Hmmm, it seems to me that this is a problem due to the difference between
noarch being a tag-like classification and packages being provided in a
tree-like classification for each architecture.
And other things, like that reason for the file system not being able to create
hard links of directo
I too thought the ports Yuri-san mentioned were not that good as examples, so
let's add these to the examples as well :)
https://www.freshports.org/ports-mgmt/pkg_tree
https://www.freshports.org/multimedia/openshot
pkg_tree is a mixture of _4 and _5.
As long as we have perl, it should work, but
Hello.
For libavif with option AOM enabled, libjxl may or may not be linked depending
on whether or not pkgconf is installed.
This is because the variable _AOM_LDFLAGS in file
${WRKSRC}/cmake/Modules/Findaom.cmake of libavif changes.
The value of _AOM_LDFLAGS comes from `pkg-config --libs aom`,
Hello.
Tijl Coosemans wrote on 2022/08/13 18:51:
> Try this patch for p11-kit. If it works you can file a bug against
> p11-kit, because I believe ports are supposed to move away from
> ca_root_nss.
>
> --- a/security/p11-kit/Makefile
> +++ b/security/p11-kit/Makefile
> @@ -25,7 +25,7 @@ MESON_AR
Hello.
I'm not sure :), but we can use pkg version -t to find out how big or small the
port version is.
pkg version -t 1.20.2_9,2 1.22.0_10,2
is <.
www/nginx-full is the sl@ve port of the m@ster port www/nginx.
So, changes in the distfile version will automatically follow.
But then the sl@ve po
Hello.
I also just recently unlinked /var/db/pkg/local.sqlite to a place where it
cannot be traced from the /var file system :)
To recover from that, there should already be a backup that others have
mentioned, in which case the following command is all that is needed to restore
it.
xzcat /var
Addition.
Ports that need to be reinstalled can usually be found in pkg check -sna.
Tatsuki Makino wrote on 2022/08/31 06:19:
> Hello.
>
> I also just recently unlinked /var/db/pkg/local.sqlite to a place where it
> cannot be traced from the /var file system :)
> To recover f
Hello.
When pkg lock/unlock with pkg-1.18.4, some packages may say "already" message.
# pkg lock pkg llvm13
pkg-1.18.4: lock this package? [y/N]: y
Locking pkg-1.18.4
llvm13-13.0.1_3: lock this package? [y/N]: y
Locking llvm13-13.0.1_3
llvm13-13.0.1_3: already locked
# pkg unlock pkg llvm13
pkg-1
Hello.
It is always present in /lib/libgcc_s.so.1.
Even a C++ program like "Hello world" will link to it on its own.
It is something we need not worry about... maybe :)
Regards.
Hello.
I cannot explain it in a long sentence in English...
Part of my understanding of the Python version and FLAVOR is that
Dan Mahoney wrote on 2022/09/30 04:52:
>
> That's ports. I'm asking about packages.
>
Currently, two versions of Python are allowed to be installed on FreeBSD.
One is
Shane Ambler wrote on 2022/09/30 13:54:
> On 30/9/22 7:05 am, Tatsuki Makino wrote:
>> Currently, two versions of Python are allowed to be installed on FreeBSD.
>> One is Python 2.7 but EOL.
>
> You can install as many python major versions as you want or need, that
> me
I tend to just keep the answer in my head and forget the why :)
And a point I remembered.
Shane Ambler wrote on 2022/10/02 12:27:
> While making or updating a port, I build and install for each flavour
> before testing with poudriere
>
> make FLAVOR=py38 stage-qa install clean
> make FLAVOR=py39
Shane Ambler wrote on 2022/10/04 15:49:
> On 3/10/22 2:06 pm, Tatsuki Makino wrote:
>> # make -C /usr/ports/multimedia/libopenshot/ install FLAVOR=py310
>> ===> py310-libopenshot-0.2.7_4 Unknown flavor 'py310', possible flavors:
>> py39.
>> *** Error co
Shane Ambler wrote on 2022/10/10 12:04:
> On 8/10/22 3:06 pm, Tatsuki Makino wrote:
>
>> This is the behavior that causes the difference between
>> USE_PYTHON=allflavors and -D BUILD_ALL_PYTHON_FLAVORS, which I am
>> relentlessly :) repeating.
>
> Notice that
Hello.
If the jail 13amd64-local-workstation for it has already been terminated, there
are other ways to try it yourself with the following command.
# Commands to be executed on the host
poudriere jail -s -j 13amd64 -p local -z workstation # -j,-p,-z are probably
for 13amd64-local-workstation :
Hello.
Naram Qashat wrote on 2022/10/13 08:46:
> https://.../data/local_aarch64-default/2022-10-10_20h05m42s/logs/errors/pkg-1.18.4.log
I am not sure about CPU ARCH that are not from Intel, but it does have
CPUTYPE=cortex-a72.
If you comment it out, does it change the behavior in any way?
I don'
Hello.
I also have blender installed.
I have it installed by a package made by poudriere, but the package build is
currently skipped because the EXAMPLES option, which is enabled by default in
graphics/opensubdiv, is BROKEN.
Until that is resolved, there will be no blender package :)
Regards.
Dima Panov wrote on 2022/10/16 05:38:
>
> EXAMPLES is off by default in opensubdiv port, please check your setup
>
Moin.
Although OPTIONS_DEFAULT is not written directly in
graphics/opensubdiv/Makefile, but is turned on by default somewhere in
Mk/bsd.options.mk.
There is a hint about that in
LuMiWa wrote on 2022/10/16 06:48:
> Difference is because you are building package but I don't. I installed
> them (pkg upgrade).
This passage is approximately equal.
port tree─┬→official poudriere─→official packages→your machine
│
└→my poudriere→my packages───→my machin
Jan Beich wrote on 2022/10/16 12:34:
>
> OPTIONS_EXCLUDE is a good way to temporarily disable broken options.
> Users don't need to re-run "make config" compared to when an option is
> moved in or out of OPTIONS_DEFAULT.
>
Before I go any further, but I cannot reproduce EXAMPLES_BROKEN=ld: error
Tatsuki Makino wrote on 2022/10/16 14:11:
> Before I go any further, but I cannot reproduce EXAMPLES_BROKEN=ld: error:...
> on my 12.3-STABLE amd64.
> What is the environment in which this can be reproduced?
> make -C /usr/ports/graphics/opensubdiv/ -D TRYBROKEN build was
Tatsuki Makino wrote on 2022/10/16 15:45:
> opensubdiv-3.5.0 package was created by poudriere bulk -T.
> However, blender-3.2.2_2 build using it failed.
> There is a line in the log that may be caused by opensubdiv.
>
I don't know how to use OpenMP properly...
In opensubdiv-3.5.
Dima Panov wrote on 2022/10/18 01:48:
> Blender have an options knob OPENMP to utilize it.
Uh, I think that is off by default, but switching it to on will most likely
eliminate OpenMP linker error.
> By the way, I'll modify opensubdiv to make openmp support optional
It would be nice to be able
Hello.
Nuno Teixeira wrote on 2022/10/26 07:14:
> I'm writing a shell script to feed poudriere testports and I need to get
Is it a hook script that installs in /usr/local/etc/poudriere.d/hooks/?
If so, I think it would be helpful to see the results of a grep like the
following.
grep -r -n -e 'r
Hello.
Andy Farkas wrote on 2022/11/02 15:26:
> I cannot build audacity because sratom fails:
> Any clues?
Probably because doxygen is installed.
It may be necessary to explicitly stop the following in
sratom-0.6.14/meson.build
if not get_option('docs').disabled()
subdir('doc')
endif
Regar
I had all ports rebuilt by portmaster -af for 12.4-STABLE but mine also stopped
at audio/sratom for the same reason :)
This can be avoided by adding the following to audio/sratom/Makefile
MESON_ARGS= -Ddocs=disabled
I have not checked specifically, but it seems that changes to pkg-plist are not
Hello.
They may have already been found and are beginning to be fixed.
But it may not be yet, so I'm writing about it here just in case.
The environment is 12.4-STABLE amd64.
In a clean environment like poudriere, python is not found and a configure
error occurs.
In an environment where various
Tatsuki Makino wrote on 2022/11/16 15:54:
> In a clean environment like poudriere, python is not found and a configure
> error occurs.
> /bin/sh: no: not found
> This is because CLANG_FORMAT is set to no when clang-format is not found.
These will be fixed in Bugzilla bug 267814.
Th
Hello.
Gleb Popov wrote on 2022/11/29 15:47:
> On Tue, Nov 29, 2022 at 12:14 AM Matthias Fechner wrote:
>> But what does the last `:` at the end of this string?
>
> This one is a correct interpretation.
>
> As for the first one, I'm also unsure about the last ":" Maybe it is a
> part of the :N
Hello.
Philipp Ost wrote on 2022/12/07 05:48:
> [ 877.223] (EE) Segmentation fault at address 0x1
I had the same segmentation fault on 12.4-STABLE amd64.
I am also a radeon. The chipset is caicos.
For now, that's all I know yet :)
Regards.
Tatsuki Makino wrote on 2022/09/13 07:51:
> When pkg lock/unlock with pkg-1.18.4, some packages may say "already" message.
> # pkg lock pkg llvm13
> llvm13-13.0.1_3: lock this package? [y/N]: y
> Locking llvm13-13.0.1_3
> llvm13-13.0.1_3: already locked
I have only seen
Tatsuki Makino wrote on 2022/12/09 17:04:
> I have only seen the log of the 1.18.99.7 change, but I assume that is fixed
> because I changed it to read all packages for which the lock is to be changed
> before changing the lock.
aah, "because it have been changed", not "because I changed it" :)
Actually, I recently created a spell to automate how to find packages that are
taking a long time to create because of ALLOW_MAKE_JOBS* :)
This is finding out what will be needed for more than 3 minutes.
#!/bin/sh
find -- /usr/local/poudriere/data/logs/bulk/latest-per-pkg/ -iname \*.log\
 -exec g
Hello.
Mel Pilgrim wrote on 2022/12/12 02:52:
> But I need to also specify a version spec. However, if I do this:
> RUN_DEPENDS= foo>=1.2:devel/foo
> That only works if the default flavor is installed. If one of the
> non-default flavors is installed, the Ports System doesn't find it and tries
Hello.
Shane Ambler wrote on 2022/12/16 13:26:
> When I build a port with poudriere I am not getting the meta and
> packagsite files to update pkg, leaving me unable to install pkgs onto
> the host system
> Any ideas why this has stopped working?
If you are using Apache for your HTTP server, it ma
Shane Ambler wrote on 2022/12/16 13:57:
> I have apache setup to view the build progress but have used file:// for
> pkg config to access the repo. This was working for several years
>
> The packagesite files are not being made after the pkgs are built, they
> are not in the .real_xxx dir
Then I
Mel Pilgrim wrote on 2022/12/17 06:49:
> On 2022-12-13 19:34, Tatsuki Makino wrote:
>> RUN_DEPENDS= foo>=1.2:devel/foo@nondefaultflavor
> No, that makes the port depend on that specific flavor. I need to allow any
> flavor to be installed.
>> The majority of the Python mo
Mel Pilgrim wrote on 2022/12/17 20:19:
> No, that's the point I've been getting at: it needs to work regardless of the
> flavor installed.
> No, that is making it flavor-specific. It just changes the dependency to
> whatever is installed at build time, not creating a dependency that will work
>
Hello.
Nuno Teixeira wrote on 2022/12/30 02:41:
> I'm looking for the best way to use `uname -m` on plist_sub:
> ---
> UNAME_M_CMD=${UNAME} -m
> PLIST_SUB+= ARCH="$$(${UNAME_M_CMD})"
> ---
> but I'm not sure sure about -m and -p differences and effects on
> arm64/aarch64.
>
> It would be
Tatsuki Makino wrote on 2022/12/30 06:17:
> For x11-wm/nscde, this is present in ${WRCSRC}/configure.ac.
>
> OS_PLUS_MACHINE_ARCH=`uname -sm | tr ' ' '_'`
> AC_SUBST(OS_PLUS_MACHINE_ARCH)
>
Then, in the case of security/zeek, it seems that un
Hello.
Ports with CARGO_CRATES=fontdb-0.10.0 (e.g. multimedia/libopenshot with option
RESVG turned on) may emit logs like the following
Warning (in fontdb:365): Failed to parse fontconfig because: IO error: No such
file or directory (os error 2)
This is due to the fact that /etc/fonts/fonts.co
Hello.
This is an unimportant topic that may or may not be useful and can be skipped :)
Today, our poudriere would have output the following
[00:01:24] [Dry Run] Deleting rust-1.66.1.pkg: missing dependency: curl-7.87.0
This causes my poudriere to spend 4 hours rebuilding rust.
Not long ago, my
Nuno Teixeira wrote on 2023/01/20 19:14:
> Other trick for those using -devel (-b latest||quarterly) is to start
> building until dependency is done, stop poudriere and restart it. Next
> build will use rust-1.66.1 from cache without rebuild it. :)
I haven't used poudriere-devel for a while becaus
Hello.
By the way, gst-plugins-bad-1.20.5/ext/zxing/meson.build has override_options :
['cpp_std=c++11'],.
What happens when it is rewritten to c++17?
Regards.
Per olof Ljungmark wrote on 2023/01/28 00:54:
> In trying update to gstreamer1-plugins-bad to 1.20.5
>
> FreeBSD 13.1-STABLE #0 stabl
Hmmm.
The following line does not seem to imply that plug-ins should not be built
even when libraries are found.
zxing_dep = dependency('zxing', version : '>= 1.1.1', required :
get_option('zxing'))
So even though the port gstreamer1-plugins-bad-1.20.5 is not supposed to be
touched with respect
Hello.
The build of www/chromium was failing in the following.
This is an excerpt from a parallel build that had other outputs mixed in.
FAILED: obj/third_party/wayland/wayland_cursor/os-compatibility.o
cc -MMD -MF obj/third_party/wayland/wayland_cursor/os-compatibility.o.d -DUSE_AU
RA=1 -DUSE_GLI
In chromium-110.0.5481.77, use_system_libwayland is switched to false.
https://cgit.freebsd.org/ports/commit/www/chromium?id=230cf3bad636e34e916bcba192aa5934d8fc5211
${WRKSRC}/third_party/wayland/src/cursor/os-compatibility.c of www/chromium and
${WRKSRC}/cursor/os-compatibility.c of graphics/way
Hello.
I was introduced to this feature in a reply to an email titled "[through-able]
poudriere: I don't want to rebuild rust with PORTREVISION bump of curl" that I
wrote on or about 2023-01-20.
This still means that the dependencies held in the package must match up to the
version number to b
chromium-110.0.5481.100 can be built and used on 12.4-STABLE amd64 with this
patch.
diff --git a/www/chromium/files/patch-third__party_wayland_include_config.h
b/www/chromium/files/patch-third__party_wayland_include_config.h
index 2dc3c949e7..df8722cf99 100644
--- a/www/chromium/files/patch-thir
Hello.
I'm a little concerned about it too, but...
> xenoveritas-abuse
> abuse-xenoveritas
I think it would be that way, and I don't think it would be that way :)
I already have a synthesizer helm port in my local tree.
However, it's name conflicts with sysutils/helm.
So I work around it by doi
Hello.
The problem stops there,
Please see the email I sent to freebsd-ports ml.
In the archives, perhaps this.
https://lists.freebsd.org/archives/freebsd-ports/2023-February/003455.html
The packages for 12.x do not even seem to have been built, as it seems to have
been made using 12.3, maybe.
Hello.
There is a list of value pairs in libpkg/private/elf_tables.h that could be
used for that, maybe?
Regards.
Hello.
I saw it in python relations.
It was due to differences between setuptools and distutils, etc.
For example, bug 265203 of bugzilla.
If not, I don't know :)
Regards.
Uh-huh? :)
You are doing something that doesn't use pkg, so it is going into
libpkg/pkg_elf.c, right?
Hello.
I'm planning on using mount -t nullfs /usr/compat /compat or something like
that, but I'm wondering if anyone else is already doing this? :)
Regards.
Hello.
Are you still saying that you are creating another program that does not rely
on the pkg command?
Then again, let's look at elf_tables.h.
It would be necessary to have such a conversion table.
static struct arch_trans machine_arch_translation[] = {
{ "x86:32", "i386" },
{
I see that the pkg for the bootstrapper has a function in
/usr/src/usr.sbin/pkg/config.c to create an ABI.
There is a function called pkg_get_myabi, which seems to be just the result of
the following command, with a colon in between. (In fact, it's a system call or
something.)
uname -s
expr `un
Hello.
Which is RAV1E option for graphics/libheif?
And Which is HEIF option for graphics/gd?
Since multimedia/librav1e pulls in cmake instead of cmake-core, it is likely to
cause that cycle.
Regards.
Hello.
I don't know yet, but there is a section in Mk/Uses/cargo.mk that appends
devel/cmake to BUILD_DEPENDS.
Although devel/cmake is a meta port and the documentation ports are separated,
it may be a loop because some graphics related ports are needed to build the
cmake documentation.
It has
Cy Schubert wrote on 2023/03/15 12:33:
>
> Are there people who still use amd (am-utils)?
>
I think it's one that will be removed if we don't say I use it, so I'll say I'm
using it :)
Regards.
Hello.
Graham Perrin wrote on 2023/03/26 14:39:
> I forgot to mention, I ran this before my first email:
> % ls -hln /usr/local/man/man3/explain_timerfd_create_or_die.3.gz
> -rw-r--r--  1 0  0   1.2K 18 May  2022
> /usr/local/man/man3/explain_timerfd_create_or_die.3.gz
> % pkg provides explain_ti
Matthew Seaman wrote on 2023/04/12 01:54:
> On 11/04/2023 12:41, Matthias Fechner wrote:
>> If I now try to remove heimdal:
>> root@gitlab:/usr/local/www/gitlab-ce # pkg delete heimdal
>> Checking integrity... done (0 conflicting)
>> Deinstallation has been requested for the following 2 packages (o
Matthias Fechner wrote on 2023/04/12 13:22:
>> pkg info -b heimdal
> heimdal-7.8.0_6:
(omit)
> Â Â Â libgssapi.so.3
(omit)
>> pkg info -B gitlab-shell
> gitlab-shell-14.18.0:
> Â Â Â libgssapi.so.3
It seems to be related here.
But I don't even know if pkg works with it as a dependency there :)
Tatsuki Makino wrote on 2023/04/12 13:29:
> It would have to be LIB_DEPENDS, not BUILD_DEPENDS.
To begin with, I would like portlint to warn us to use USES=gssapi:heimdal :)
bob prohaska wrote on 2023/04/13 11:18:
> Running www/chromium (from ports via poudriere) on a Pi4
> reports a stream of
> [30033:1100602368:0412/190147.848068:ERROR:bus.cc(399)] Failed to connect to
> the bus: Could not parse server address: Unknown address type (examples of
> valid types are
Gleb Popov wrote on 2023/04/13 15:08:
>> That is because there is a setenv("DBUS_SESSION_BUS_ADDRESS","disabled:"
>> somewhere :)
>
> Most likely, this is
> https://bugs.chromium.org/p/chromium/issues/detail?id=918234
>
>> The workaround is DBUS_SESSION_BUS_ADDRESS below.
> DBUS_SESSION_BUS_ADDR
Hello.
This is an unimportant topic that may or may not be useful and can be skipped :)
Tatsuki Makino wrote on 2023/01/20 17:40:
> poudriere: I don't want to rebuild rust with PORTREVISION bump of curl
Some of the same kind of stories include :)
I don't want to rebuild gcc12 w
Hello.
It appears that INDEX-12 (DESCRIBE.12) has not been updated by portsnap for
several days.
For example, gstreamer1 remains in the following version
> zgrep -e ^gstreamer1-1 /var/db/portsnap/files/e24b4779*a78a1f31.gz | cut -f 1
> -d \|
gstreamer1-1.22.0_1
Is it under some kind of mainten
Colin Percival wrote on 2023/05/04 15:42:
> A recent commit (be67c1c353ca) broke Portsnap's INDEX generation; somehow
I understood what was going on.
Thank you very much.
I will adopt a different approach for a while. pkg version -P ... , make -C
/usr/ports fetchindex , or others.
Regards.
Hello.
texlive-base-20230313_2 has library dependency on poppler-23.04.0.
However, I cannot find the basis for this by running the following command.
pkg info -lq texlive-base-20230313_2 | xargs readelf -d -- | grep libpoppler
pkg info -lq texlive-base-20230313_2 | xargs strings -- | grep poppler
locate.database is created by /etc/periodic/weekly/310.locate, isn't it?
It's behavior is run by nobody.
The other thing that is often forgotten is that the group behaves as if a
setgid bit has been set up in the directory.
Is the idea that one daemon-specific user is sufficient, as long as sete
Hello.
I often use make makepatch because it is very useful, but under certain
conditions, new patches are not written out to PATCHDIR directory.
The condition may be patched such as
/usr/ports/www/firefox/files/patch-libwebrtc-generated.
This patch includes patches for the following files
---
Hello.
Now that there seems to be people who are building chromium and have very fast
machines, I'll hang around :)
I am building chromium using a 4 threads cpu and limiting MAKE_JOBS_NUMBER to 3.
However, there seem to be moments when the load average rises to about 7, even
though nothing else
Hello.
Here are the commands in my notes.
The prompt on the far left is whether to run on the host or jail side.
The name I am giving to the poudriere jail is src.
host# poudriere bulk -j src -i x11/xauth
jail# make -C /usr/ports/x11/xterm/ pkg-depends install-package
jail# make -C /usr/ports/dev
Hello.
Nuno Teixeira wrote on 2023/05/30 19:53:
> Thats a big difference on compilation times:
>
> ---
> build of www/firefox | firefox-113.0.2_1,2
> build time: 05:35:57
> ---
> build of www/chromium | chromium-113.0.5672.126
> build time: 56:59:23
> ---
The ratio is roughly similar to that for
Hello.
Someone said something similar to that. :)
-> https://lists.freebsd.org/archives/freebsd-ports/2023-January/003292.html
It seems that the package download feature of poudriere-devel has to keep pace
with the official builder.
Perhaps when updates are made to chromium, etc., the official w
Hello.
> PORTSDIR= /var/ports/msrvkx
> vm.tfc:/ds/ports/freebsd/d830ac4c7cfd
> 9.9T129G9.8T 1%/usr/ports
> root@vm:~ # rm -rf /ds/ports/freebsd/head> root@vm:~ # git clone
> https://github.com/freebsd/freebsd-ports.git /ds/
Hello.
Mark Millard wrote on 2023/07/05 06:51:
> [I continued to type MAX_JOBS_NUMBER where
> MAKE_JOBS_NUMBER should have been what I
> typed.]
>
It may be possible to set stricter restrictions on selected ports in
/usr/local/etc/poudriere.d/make.conf.
For example,
# normally
.if ${.CURDIR:tA
Hello.
These days, when we are using graphical, llvm gets installed by runtime
dependencies, so it is not that much of a load to depend on it at build time.
And, likewise, these days there are variables such as BINARY_ALIAS, so there is
no need for unified diff patches for that reason.
As expec
Hello.
Kevin Oberman wrote on 2023/07/10 09:52:
> only completed 459 packages, a rate of 10 packages/hour and, if it
> continues at this rate, will not complete for about 40 days. Nt sure if
The numbers in that area are meaningless :)
For example, the following two
poudriere bulk -j ... -C devel
Kevin Oberman wrote on 2023/07/10 14:01:
> I have been monitoring the builds of amd64 builds for at least two years
> and this is way outside of the norm. It continues to build packages at a
> rate of 10 per hour which would still mean that the build will not complete
> until late August. I guess
Hello.
Guido Falsi wrote on 2023/07/10 18:01:
> This resonates with something I have observed from time to time on my build
> machine.
> Especially when using a big ccache (32 GiB or more, is what I can testify
> about), but also when building ports with a big number of files like the
> teTeX,
Jan Beich wrote on 2023/07/11 12:51:
> Even with pkg-1.20.2 many running jobs are stuck with *-depends in Status.
>
> https://pkg-status.freebsd.org/beefy18/build.html?mastername=main-amd64-default&build=pd6fe15e0ee7a_s1ec7cb1b5d
At the time I started writing this email, several qt5-* and kf5-*s
Hello.
Perhaps this.
=>> Building audio/libopenshot-audio
　　　︙
Host OSVERSION: 1400089
Jail OSVERSION: 1204000
It happens sometimes :)
Regards.
Guido Falsi wrote on 2023/07/11 16:27:
> Anyway you can monitor "vfs.freevnodes" and "vfs.numvnodes" to discover if
> that's an issue for you. If numvnodes is very near or above maxvnodes nad at
> the same time freevnodes is very low you have a problem. Another indicator
> could be vfs.vnodes_cr
Tatsuki Makino wrote on 2023/07/10 11:13:
> Kevin Oberman wrote on 2023/07/10 09:52:
>> only completed 459 packages, a rate of 10 packages/hour and, if it
>> continues at this rate, will not complete for about 40 days. Nt sure if
>
> The numbers in that area are meanin
Hello.
Graham Perrin wrote on 2023/08/05 03:33:
> On 04/08/2023 15:43, void wrote:
An example of just a maintainer case hangs here.
All ports which I am a maintainer use the same email address as this email.
The email address I use for my FreeBSD Bugzilla account is the same as this
email.
And
1 - 100 of 333 matches
Mail list logo
