On Wed, Mar 12, 2014 at 11:15:43AM -0600, Eric Blake wrote: > Revisiting an older thread > > On 11/26/2013 07:38 AM, Martin Kletzander wrote: > > On Tue, Nov 26, 2013 at 10:14:36AM +0000, Daniel P. Berrange wrote: > >> On Tue, Nov 26, 2013 at 10:58:25AM +0100, Martin Kletzander wrote: > >>> Adding a support for LIBVIRT_API_PATH evironment variable, which can > >>> control where the script should look for the 'libvirt-api.xml' file. > >>> This allows building libvirt-python against different libvirt than the > >>> one installed in the system. This may be used for example in autotest > >>> or by packagers without the need to install libvirt into the system. > >>> > > >>> - libvirt_api = get_pkgconfig_data(["--variable", "libvirt_api"], "libvirt") > >>> + libvirt_api = os.getenv("LIBVIRT_API_PATH") > >>> + > > >> > >> NACK, setting pkg-config already takes care of this. See the > >> build-many.sh scrpit attached to this mail which demonstrates > >> use of PKG_CONFIG_PATH to build against every version of libvirt > >> back to 0.9.11 > >> > > > > This still means you have to configure libvirt with different prefix, > > install it and then you can use PKG_CONFIG_PATH. This variable (which > > is unused if unset) makes it easier to use in case you have it built > > with default prefix etc. It would help me a lot, but if everyone else > > is OK with installing libvirt in order to build python bindings just > > to test something, I'll keep this in my git. > > I'm still interested in the ability to test libvirt-python against an > uninstalled libvirt tree. Should we revisit this patch, or something > like it? > I'd be still happy to have this in, especially for us developers. It's useful when testing a libvirt-python patch against upstream libvirt which, in this case, doesn't have to be installed to make the bindings work. Martin
Attachment:
signature.asc
Description: Digital signature
-- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list