We use a LP build recipe that includes livecd-rootfs.
We are moving from bzr to git due to other projects moving to git.
When we import livecd-rootfs into git it does not commit empty
directories into the git tree. The livecd-build/ubuntu-
cpc/includes.chroot/etc/network/interfaces.d/ directory is dropped
when we do this import.
[Impact]
* Importing the package and committing it to a git repository
drops empty directories from the ubuntu-cpc/includes.chroot tree
* Images built in the ubuntu-cpc project drop this directory which
is a regression
[Test Case]
* Create a bare git repo,
* Pull the livecd-rootfs from xenial-updates into the repo,
* Commit the code to git
* Use the committed code (fresh clone) to build the ubuntu-cpc project
* Inspect the images to see if /etc/network/interfaces.d exists
[Regression Potential]
* Low. This create a directory if it does or doesn't exist so builds from
sources that include the empty directory should not be impacted. The
difference would be owner or permissions between the two methods (includes.chroot
and the chroot hook) and my testing with this patch has shown that to
be correct.
We use a LP build recipe that includes livecd-rootfs. build/ubuntu- chroot/ etc/network/ interfaces. d/ directory is dropped
We are moving from bzr to git due to other projects moving to git.
When we import livecd-rootfs into git it does not commit empty
directories into the git tree. The livecd-
cpc/includes.
when we do this import.
[Impact]
* Importing the package and committing it to a git repository cpc/includes. chroot tree
drops empty directories from the ubuntu-
* Images built in the ubuntu-cpc project drop this directory which
is a regression
[Test Case]
* Create a bare git repo, interfaces. d exists
* Pull the livecd-rootfs from xenial-updates into the repo,
* Commit the code to git
* Use the committed code (fresh clone) to build the ubuntu-cpc project
* Inspect the images to see if /etc/network/
[Regression Potential]
* Low. This create a directory if it does or doesn't exist so builds from
sources that include the empty directory should not be impacted. The
difference would be owner or permissions between the two methods (includes.chroot
and the chroot hook) and my testing with this patch has shown that to
be correct.