On 05/30/2012 03:02 PM, Eric Blake wrote:
On 05/30/2012 06:43 AM, Eric Blake wrote:
On 05/30/2012 04:15 AM, Martin Kletzander wrote:
I added libvirt_qemu_probes.h into BUILT_SOURCES. That makes it generated, but most probably it is not the clearest way how to do that, but it fixes the build. --- Feel free to clean this, but for the time being, I'm pushing this as a build breaker.
src/Makefile.am | 1 + 1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/src/Makefile.am b/src/Makefile.am index c8efa5b..1fde936 100644 --- a/src/Makefile.am +++ b/src/Makefile.am @@ -1294,6 +1294,7 @@ BUILT_SOURCES += libvirt_probes.h libvirt_probes.stp libvirt_functions.stp if WITH_QEMU libvirt_driver_qemu_la_LIBADD += libvirt_qemu_probes.lo nodist_libvirt_driver_qemu_la_SOURCES = libvirt_qemu_probes.h +BUILT_SOURCES += libvirt_qemu_probes.h endif
You're the third person to post this; I still think my patchset (pending review) tackles the problem a bit better: https://www.redhat.com/archives/libvir-list/2012-May/msg01301.html https://www.redhat.com/archives/libvir-list/2012-May/msg01308.html
Actually, on re-reading my patchset, I was tackling a different issue for VPATH builds. I think your patch is correct after all, just that we still need my patches as well once they get reviewed. BUILT_SOURCES does not cause the file to be distributed, but does cause it to be built earlier in the process, which is what we need since compilation of other files depends on it existing.
I forgot to mention in the patch, that this fixes the build when dtrace is enabled. Without that, it works fine. I'm not sure where to put it, the rule is there, but the dependency for the header file is missing. I have also problem with make check, but that's a problem of missing rule for the augeas build of sanlock config file and I didn't have the time (and will) to look at it yet. Maybe it will be fixed by your patchset, I'll go through it asap. Martin