[OE-core] [PATCH 1/1] perf: split package

Bruce Ashfield bruce.ashfield at gmail.com
Thu Jul 17 13:22:58 UTC 2014


On Thu, Jul 17, 2014 at 2:58 AM, Chong Lu <Chong.Lu at windriver.com> wrote:
>
> On 07/17/2014 10:13 AM, Bruce Ashfield wrote:
>>
>> On Wed, Jul 16, 2014 at 9:43 PM, Chong Lu <Chong.Lu at windriver.com> wrote:
>>>
>>> On 07/16/2014 09:04 PM, Bruce Ashfield wrote:
>>>>
>>>> On Wed, Jul 16, 2014 at 4:00 AM, Chong Lu <Chong.Lu at windriver.com>
>>>> wrote:
>>>>>
>>>>> Fix incorrect paths to split perf package.
>>>>
>>>> Can you elaborate here ? This commit message doesn't really tell us
>>>> much.
>>>> Why is the current split incorrect ? What is the symptom and result of
>>>> the
>>>> incorrect split ? Are the package contents the same after this change ?
>>>>
>>>> Bruce
>>>
>>> Hi Bruce,
>>>
>>> Currently, perf can't split to perf-archive, perf-tests, perf-python and
>>> perf-perl.
>>> All files are included in perf package. For example, like perf-tests
>>> package:
>>> It generates files in '${exec_prefix}/libexec/perf-core/tests', not in
>>> '${libdir}/perf/perf-core/tests'.
>>> So I change the files paths to make split successfully.
>>
>> But that means, whether or not we like it, anything using perf has been
>> using
>> those "non splits" and getting everything when perf is installed.
>>
>> It's also might be in order to make the main perf package RDEPEND on
>> the other three, to keep it consistent. Make it a ?= so it can be
>> overridden,
>> and then we can age it out over time.
>>
>> Bruce
>
> Do you mean we should set RDEPENDS like:
>     RDEPENDS_${PN} ?= "${PN}-archive ${PN}-tests ${PN}-python ${PN}-perl"
>
> In perf.bb file, we have RDEPENDS_${PN} += "elfutils", so I think we should
> use
>     RDEPENDS_${PN} += "${PN}-archive ${PN}-tests ${PN}-python ${PN}-perl"
> to keep it consistent. Do you agree?

Consistency is one thing, but it won't allow the split package to be
easily dropped,
so I'd still suggest that the ?= would make it easier to override.

Bruce

>
> Or, we can also revert following commit to make all files in perf package.
>     commit 36c3e923df85e685b707fe6bd84b053a5e366fe4
>     Author: Henning Heinold <henning at itconsulting-heinold.de>
>     Date:   Mon Jun 9 23:21:29 2014 +0200
>
>         perf: split packging
>
> Best Regards
> Chong
>
>
>>> I will improve commit message and resend a V2.
>>>
>>> Best Regards
>>> Chong
>>>
>>>>> Signed-off-by: Chong Lu <Chong.Lu at windriver.com>
>>>>> ---
>>>>>    meta/recipes-kernel/perf/perf.bb | 8 ++++----
>>>>>    1 file changed, 4 insertions(+), 4 deletions(-)
>>>>>
>>>>> diff --git a/meta/recipes-kernel/perf/perf.bb
>>>>> b/meta/recipes-kernel/perf/perf.bb
>>>>> index 20968f9..781a738 100644
>>>>> --- a/meta/recipes-kernel/perf/perf.bb
>>>>> +++ b/meta/recipes-kernel/perf/perf.bb
>>>>> @@ -166,10 +166,10 @@ RSUGGESTS_${PN} += "${PN}-archive ${PN}-tests
>>>>> ${RSUGGESTS_SCRIPTING}"
>>>>>
>>>>>    FILES_${PN} += "${libexecdir}/perf-core
>>>>> ${exec_prefix}/libexec/perf-core ${libdir}/traceevent"
>>>>>    FILES_${PN}-dbg += "${libdir}/python*/site-packages/.debug"
>>>>> -FILES_${PN}-archive = "${libdir}/perf/perf-core/perf-archive"
>>>>> -FILES_${PN}-tests = "${libdir}/perf/perf-core/tests"
>>>>> -FILES_${PN}-python = "${libdir}/python*/site-packages
>>>>> ${libdir}/perf/perf-core/scripts/python"
>>>>> -FILES_${PN}-perl = "${libdir}/perf/perf-core/scripts/perl"
>>>>> +FILES_${PN}-archive = "${exec_prefix}/libexec/perf-core/perf-archive"
>>>>> +FILES_${PN}-tests = "${exec_prefix}/libexec/perf-core/tests"
>>>>> +FILES_${PN}-python = "${libdir}/python*/site-packages
>>>>> ${exec_prefix}/libexec/perf-core/scripts/python"
>>>>> +FILES_${PN}-perl = "${exec_prefix}/libexec/perf-core/scripts/perl"
>>>>>
>>>>>
>>>>>    INHIBIT_PACKAGE_DEBUG_SPLIT="1"
>>>>> --
>>>>> 1.9.1
>>>>>
>>>>> --
>>>>> _______________________________________________
>>>>> Openembedded-core mailing list
>>>>> Openembedded-core at lists.openembedded.org
>>>>> http://lists.openembedded.org/mailman/listinfo/openembedded-core
>>>>
>>>>
>>>>
>>
>>
>



-- 
"Thou shalt not follow the NULL pointer, for chaos and madness await
thee at its end"



More information about the Openembedded-core mailing list