Re: [PATCH 1/1] gst-plugins-bad: add curl as build dependency

[Richard Purdie <richard.purdie@linuxfoundation.org>]

On Mon, 2012-04-16 at 16:40 +0200, Andreas Oberritter wrote:
> On 15.04.2012 01:47, Xu, Dongxiao wrote:
> > On Sat, 2012-04-14 at 13:23 +0200, Koen Kooi wrote:
> >> Op 14 apr. 2012, om 12:44 heeft Dongxiao Xu het volgende geschreven:
> >>
> >>> gst-plugins-bad needs curl.h to finish its build, so add curl as its
> >>> dependency.
> >>
> >> I can't be sure without the exact failure log, but: missing PR bump
> > 
> > Add a dependency here doesn't impact the original build. If the original
> > build is successful, it means curl has already been built by chance. Do
> > we still need to bump PR in this case?
> > 
> > Besides, this is the detailed log:
> > 
> > i586-poky-linux-libtool: compile:  i586-poky-linux-gcc -m32 -march=i586
> > --sysroot=/distro/sdb/build-master/tmp/sysroots/qemux86 -std=gnu99
> > -DHAVE_CONFIG_H -I. -I../.. -pthread
> > -I/distro/sdb/build-master/tmp/sysroots/qemux86/usr/include/gstreamer-0.10 -I/distro/sdb/build-master/tmp/sysroots/qemux86/usr/include/glib-2.0 -I/distro/sdb/build-master/tmp/sysroots/qemux86/usr/lib/glib-2.0/include -I/distro/sdb/build-master/tmp/sysroots/qemux86/usr/include/libxml2 -pthread -I/distro/sdb/build-master/tmp/sysroots/qemux86/usr/include/gstreamer-0.10 -I/distro/sdb/build-master/tmp/sysroots/qemux86/usr/include/glib-2.0 -I/distro/sdb/build-master/tmp/sysroots/qemux86/usr/lib/glib-2.0/include -I/distro/sdb/build-master/tmp/sysroots/qemux86/usr/include/libxml2 -DG_THREADS_MANDATORY -DG_DISABLE_CAST_CHECKS -DG_DISABLE_ASSERT -Wall -Wdeclaration-after-statement -Wvla -Wpointer-arith -Wmissing-declarations -Wmissing-prototypes -Wredundant-decls -Wwrite-strings -Wformat-security -Wold-style-definition -Winit-self -Wmissing-include-dirs -Waddress -Wno-multichar -Wnested-externs -O2 -pipe -g -feliminate-unused-debug-types -c gstcurlsink.c  -fPIC -DPIC -o .libs/libg!
 stc
>  ur
> >  l_la-gstcurlsink.o
> > i586-poky-linux-libtool: compile:  i586-poky-linux-gcc -m32 -march=i586
> > --sysroot=/distro/sdb/build-master/tmp/sysroots/qemux86 -std=gnu99
> > -DHAVE_CONFIG_H -I. -I../.. -pthread
> > -I/distro/sdb/build-master/tmp/sysroots/qemux86/usr/include/gstreamer-0.10 -I/distro/sdb/build-master/tmp/sysroots/qemux86/usr/include/glib-2.0 -I/distro/sdb/build-master/tmp/sysroots/qemux86/usr/lib/glib-2.0/include -I/distro/sdb/build-master/tmp/sysroots/qemux86/usr/include/libxml2 -pthread -I/distro/sdb/build-master/tmp/sysroots/qemux86/usr/include/gstreamer-0.10 -I/distro/sdb/build-master/tmp/sysroots/qemux86/usr/include/glib-2.0 -I/distro/sdb/build-master/tmp/sysroots/qemux86/usr/lib/glib-2.0/include -I/distro/sdb/build-master/tmp/sysroots/qemux86/usr/include/libxml2 -DG_THREADS_MANDATORY -DG_DISABLE_CAST_CHECKS -DG_DISABLE_ASSERT -Wall -Wdeclaration-after-statement -Wvla -Wpointer-arith -Wmissing-declarations -Wmissing-prototypes -Wredundant-decls -Wwrite-strings -Wformat-security -Wold-style-definition -Winit-self -Wmissing-include-dirs -Waddress -Wno-multichar -Wnested-externs -O2 -pipe -g -feliminate-unused-debug-types -c gstcurl.c  -fPIC -DPIC -o .libs/libgstcu!
 rl_
>  la
> >  -gstcurl.o
> > gstcurlsink.c:45:23: fatal error: curl/curl.h: No such file or directory
> > compilation terminated.
> > make[3]: *** [libgstcurl_la-gstcurlsink.lo] Error 1
> > make[3]: *** Waiting for unfinished jobs....
> 
> That's certainly wrong, because for me curl is a new dependency, i.e. it
> didn't need to be built before this patch went in.
> 
> The error above is probably the symptom of a race condition, where curl
> was rebuilt and thus removed from the sysroot after gst-plugins-bad was
> configured.
> 
> So either a PR bump is required or --disable-curl (or such) should have
> been used instead.

I want the builds to be deterministic so either enabling curl (via
DEPENDS) or disabling it through the configure option would be fine. I
went with adding curl since gst-plugins-bad is one of the last packages
to build on the autobuilder and hence it would usually have had curl
built so that is the most tested configuration I'm aware of.

I've pushed a PR bump.

Cheers,

Richard