Tree/Branch: Poky/Master Commit: d4132fa12885fc050313a5c9aa6903e4fa92c94f Not sure if it's the configuration problem on my machine. Just keep it here and I will try on other machine tomorrow. I find even if the image type(saying core-image-minimal) is build finished(hob shows "build completed"), there is no rootfs image generated under tmp/deploy/images. I think if the bug is valid, it shoulde be major or block issue because the output of a build is not correct.
I think I know what's going on here. There is a known issue with the image creator that the native tools for rootfs construction are not correctly included in the dependencies of the image and therefore if these native tools have not already been built the actual rootfs construction will fail. I assume this was a clean build directory?
I guess the reason why no rootfs is generated is when calling the bitbake server command 'buildTargets', hob passes in all the packages included in the core-image-minimal, not the core-image-minimal itself.
If we build a image from bitbake command line first. Then build the same image in hob. There will be no rootfs image generated also.
Currently the actual image build is queued to be run using buildFile after the package builds (using buildTargets) have completed. That's what works on my Fedora 15 workstation(s) but will fail if the builddir hasn't had an image built in it already. Fixing hob to use buildTargets for the actual image is a high priority issue for M2 but if that's not what you're seeing I'd like to get some more info so that I can try and ensure the image creator works for you. Can you tell me Python version, distribution and distro version so I can try and reproduce?
(In reply to comment #4) > Currently the actual image build is queued to be run using buildFile after the > package builds (using buildTargets) have completed. > That's what works on my Fedora 15 workstation(s) but will fail if the builddir > hasn't had an image built in it already. > Fixing hob to use buildTargets for the actual image is a high priority issue > for M2 but if that's not what you're seeing I'd like to get some more info so > that I can try and ensure the image creator works for you. > Can you tell me Python version, distribution and distro version so I can try > and reproduce? I am using 32bit Ubuntu 10.04 with python 2.6.5.
got 3 event after buildFile command was sent out: <bb.command.CommandCompleted> <bb.command.CommandFailed> <bb.command.CommandFailed> (In reply to comment #4) > Currently the actual image build is queued to be run using buildFile after the > package builds (using buildTargets) have completed. > > That's what works on my Fedora 15 workstation(s) but will fail if the builddir > hasn't had an image built in it already. > > Fixing hob to use buildTargets for the actual image is a high priority issue > for M2 but if that's not what you're seeing I'd like to get some more info so > that I can try and ensure the image creator works for you. > > Can you tell me Python version, distribution and distro version so I can try > and reproduce?
OK, working on this.
For those following along, the problem was that the GUI was calling the findConfigFilePath command of the cooker looking for a UI specific config file. However the findConfigFilePath command was buggy such that it would fire the ConfigFilePathFound event regardless of whether a file was actually found or not, leading to the UI trying to handle the event and process a None object as a file path. The submitted patch updates the cooker to only fire events in the finder methods when something is actually found.
Dammit, comment on the wrong bug. Sorry. Please ignore last comment.
This change is pending further development on #1249
http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=2d608f837d51c900dacadeb9737fbde068578d8a
Verify the bug with commit 46cf540e63a848512617b20fd8492f81bfb2f704, now rootfs image could be built out with hob.