Bug 5208

Summary: Failed to remove packages via PACKAGE_EXCLUDE
Product: [Build System, Metadata & Runtime] OE-Core Reporter: Wenzong.Fan <wenzong.fan>
Component: coreAssignee: Mark Hatle <mark.hatle>
Status: RESOLVED INVALID QA Contact:
Severity: normal    
Priority: Medium CC: bluelightning, jessica.zhang, meta.mr.watcher, meta.watcher
Version: 1.5   
Target Milestone: 1.5   
Hardware: x86   
OS: Multiple   
Whiteboard:
OS type for building Yocto: --- Type of Regression: ---
Verified: Documentation change: Don't know

Description Wenzong.Fan 2013-09-17 06:41:43 UTC 
This is a common issue that not only for specfic packages, remove python-smartpm for example:

1) echo echo PACKAGE_EXCLUDE = "python-smartpm"  >> local.conf
2) bitbake core-image-sato

Errors:
| Note: adding Smart RPM DB channel
|
| Note: to be installed:  packagegroup-core-ssh-dropbear@all packagegroup-core-x11-sato-games@qemux86 packagegroup-core-x11-base@all smartpm@i586 psplash@i586 packagegroup-core-boot@qemux86 packagegroup-base-extended@qemux86 rpm@i586 packagegroup-core-x11-sato@qemux86 locale-base-en-us@i586 locale-base-en-gb@i586
| Loading cache...
| Updating cache...               ######################################## [100%]
|
| Computing transaction...error: Can't install smartpm-1.4.1-r9@i586: unable to install provider for python-smartpm:
| error:     python-smartpm-1.4.1-r9@i586 is to be excluded
|
| Saving cache...
|
| WARNING: exit code 1 from a shell command.

Removing perl, libpam ... would have the same issue.

We should work out a mechanism to remove all the rpms that generated by an excluded package, not only the main rpms for it.
Comment 1 Paul Eggleton 2013-09-17 13:55:24 UTC 
PACKAGE_EXCLUDE is a list of packages, not recipes, and each package corresponds to a single rpm. It sounds like you want it to be a recipe exclusion but I don't think that is really practical with the way this feature has been implemented.

FWIW, an aside perhaps, but I've always thought the split between "python-smartpm" and "smartpm" isn't particularly useful, so I wouldn't necessarily point to this as a good example.
Comment 2 Paul Eggleton 2013-09-17 16:21:18 UTC 
After some discussion with RP/Saul I'm marking this as invalid based on the above.