[OE-core] [PATCH][RFC] insane.bbclass: Enhance file-rdeps QA check
Alejandro Hernandez
alejandro.hernandez at linux.intel.com
Thu Jan 29 21:20:06 UTC 2015
On 29/01/15 08:18, Saul Wold wrote:
> Have you done before and after builds to assess the performance
> differences if any?
I checked, found no actual difference on performance.
> What tool are you using that might cause double spacing in your commit
> messages.
Won't be happening again.
>
>> diff --git a/meta/classes/insane.bbclass b/meta/classes/insane.bbclass
>> index 143ec46..542346a 100644
>> --- a/meta/classes/insane.bbclass
>> +++ b/meta/classes/insane.bbclass
>> @@ -34,7 +34,7 @@ WARN_QA ?= "ldflags useless-rpaths rpaths staticdev
>> libdir xorg-driver-abi \
>> ERROR_QA ?= "dev-so debug-deps dev-deps debug-files arch pkgconfig
>> la \
>> perms dep-cmp pkgvarcheck perm-config perm-line
>> perm-link \
>> split-strip packages-list pkgv-undefined var-undefined \
>> - version-going-backwards \
>> + version-going-backwards symlink-to-sysroot\
>> "
>>
> We should also add this to the poky.conf, for the Poky distro since it
> overrides the ERROR_QA setting in meta-yocto/distro/conf/poky.conf.
> That patch should go to poky at yoctoproject.org
Yes, a patch to poky is also required, I intended to send it once this
gets accepted.
>> @@ -841,31 +841,27 @@ def package_qa_check_rdepends(pkg, pkgdest,
>> skip, taskdeps, packages, d):
>> # case there is a RDEPENDS_pkg = "python" in the
>> recipe.
>> for py in [ d.getVar('MLPREFIX', True) + "python",
>> "python" ]:
>> if py in done:
>> - filerdepends.discard("/usr/bin/python")
>> + filerdepends.pop("/usr/bin/python",None)
>> done.remove(py)
>> for rdep in done:
>> # For Saving the FILERPROVIDES, RPROVIDES and
>> FILES_INFO
>> - rdep_rprovides = set()
>> rdep_data =
>> oe.packagedata.read_subpkgdata(rdep, d)
>> for key in rdep_data:
>> if key.startswith("FILERPROVIDES_") or
>> key.startswith("RPROVIDES_"):
>> for subkey in rdep_data[key].split():
>> - rdep_rprovides.add(subkey)
>> + filerdepends.pop(subkey,None)
> Should this be a pop here since you are removing a add()?
Yes, it should be a pop, because it was adding subkey to rdep_rprovides,
which were substracted from filerdepends afterwards, this worked on sets
but not on dictionaries, the same result is achieved this way.
More information about the Openembedded-core
mailing list