x11/nvidia-settings: poudriere fails; portmaster succeeds

[David Wolfskill]

In fairness, this may be an "apple vs. orange" comparison.  But it's
fairly unusual (in my experience) for poudriere to fail to build a port,
but when it's a port that I had just built successfully (using
portmaster) on my laptop... well, I thought it was worth mentioning.

The whine in the poudriere log is:

...
cc -O2 -pipe  -fstack-protector -fno-strict-aliasing -fno-strict-aliasing -fno-omit-frame-pointer -Wformat=2 -Wno-unused-parameter -Wno-format-zero-length -DNV_BSD  -I/usr/local/include  -I . -I image_data -I libXNVCtrl -I XF86Config-parser/.. -I libXNVCtrlAttributes -I xpm_data -I common-utils -I common-unix/virtual-resolutions -I _out/FreeBSD_amd64 -I/usr/local/include -D_THREAD_SAFE -pthread  -I /usr/include/dbus-1.0/ -DPROGRAM_NAME=\"nvidia-settings\" -fPIC -c XF86Config-parser/Generate.c -o _out/FreeBSD_amd64/Generate.o && cc -MM -O2 -pipe  -fstack-protector -fno-strict-aliasing -fno-strict-aliasing -fno-omit-frame-pointer -Wformat=2 -Wno-unused-parameter -Wno-format-zero-length -DNV_BSD  -I/usr/local/include  -I . -I image_data -I libXNVCtrl -I XF86Config-parser/.. -I libXNVCtrlAttributes -I xpm_data -I common-utils -I common-unix/virtual-resolutions -I _out/FreeBSD_amd64 -I/usr/local/include -D_THREAD_SAFE -pthread  -I /usr/include/dbus-1.0/ -DPROGRAM_NAME=\"nvidia-settings\" -fPIC XF86Config-parser/Generate.c | sed -e "s,: ,: $\(wildcard ," -e "s,\([^\\]\)$,\1)," -e "s;^Generate.o: ; _out/FreeBSD_amd64/Generate.o: ;" > _out/FreeBSD_amd64/Generate.d
gtk+-2.x/ctkgridlicense.c:42:10: fatal error: 'dbus/dbus.h' file not found
#include <dbus/dbus.h>
         ^~~~~~~~~~~~~
1 error generated.
....

which looks to me as if somehow nvidia-settings has a dependency on dbus
of which poudriere was unaware; more on that below (after I sketch the
environment).

I have two machines that have nightly-updated private mirrors of the
FreeBSD SVN src, ports, and doc repositories: my laptop and a "build
machine" ("freebeast").

As described in <http://www.catwhisker.org/~david/FreeBSD/upgrade.html>,
I perform a source-based update of stable/11 on each of the machines
each morning; upon reboot and after "make delete-old-libs", I use
portmaster on each of these machines to update the installed ports to
match the state of the just-updated /usr/ports working copy.

The laptop only builds the world & its own custom kernel; freebeast runs
a GENERIC kernel and builds that, as well as kernels for the two
"production" machines.  On a regular basis (normally, each Sunday
morning), I update the production machines by installing the just-built
snapshot of stable/11 onto them, reboot, perform a "pkg updgrade", and
reboot again.

The production machines are configured to get their packages (for the
"pkg upgrade") from freebeast.

freebeast uses poudriere to build the packages.

Since I (normally) only update the production machines on Sunday, I
don't see much value in running poudriere every day.  On the other hand,
if I wait until Sunday (to get a week's worth of updates), that can take
a while, even with a reasonably fast machine.

So, as a compromise, I do the "weekly" package-building in two stages:
the first is on Saturday (as in, today), doing a "catch up" run for the
last 6 days of updates, then on Sunday -- usually! -- there's only a
small amount of work to get caught up.

Thus, the successful build of x11/nvidia-settings was on my laptop,
yesterday:

...
===>>> The following actions were performed:
        Upgrade of nvidia-xconfig-378.13 to nvidia-xconfig-384.59
        Upgrade of libuv-1.13.1 to libuv-1.14.0
        Upgrade of cups-filters-1.13.5 to cups-filters-1.16.0
        Upgrade of harfbuzz-1.4.7 to harfbuzz-1.4.8
        Upgrade of harfbuzz-icu-1.4.7 to harfbuzz-icu-1.4.8
        Upgrade of opusfile-0.8 to opusfile-0.9
        Upgrade of nvidia-settings-378.13_1 to nvidia-settings-384.59_1

Command exit status: 0

freebeast runs (essentially) headless; it has no X11-related ports
installed at all.  So my next attempt to build x11/nvidia-settings
was on freebeast, but using poudriere -- which failed as described
above.

Checking (on my laptop) for ports on which nvidia-settings depends, I
see:

g1-252(11.1-S)[1] pkg info -d x11/nvidia-settings
nvidia-settings-384.59_1:
        libXxf86vm-1.1.4_1
        libXv-1.0.11,1
        libXext-1.3.3_1,1
        libX11-1.6.5,1
        pango-1.40.6
        gtk2-2.24.31
        fontconfig-2.12.1,1
        freetype2-2.8
        libvdpau-1.1.1
        mesa-libs-17.1.5
        gdk-pixbuf2-2.36.6
        cairo-1.14.8_1,2
        jansson-2.10
        glib-2.50.2_4,1
        gettext-runtime-0.19.8.1_1
        atk-2.24.0
g1-252(11.1-S)[2] 

And that list does not seem to mention any dbus-related ports.

So I *suspect* that the portmaster build was "successful" only by
accident (because portmaster builds in teh local enviornment, which
happened to already have dbus installed, so dbus/dbus.h was already
present -- even though not called out as a dependency), while poudriere
would (I expect) have provided the file if it had actually been advised
that it was wanted....

Sorry for going on at such lengh; hope it was of some use despite that.

Peace,
david
-- 
David H. Wolfskill				david at catwhisker.org
If we wish to eliminate sources of Fake News, start at the top: D. Trump.

See http://www.catwhisker.org/~david/publickey.gpg for my public key.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 603 bytes
Desc: not available
URL: <http://lists.freebsd.org/pipermail/freebsd-ports/attachments/20170819/31c71c71/attachment.sig>