[OE-core] [PATCH v2] recipes-core: move hwclock.sh to util-linux

[Alex Stewart]

On 01/13/2018 05:48 AM, Otavio Salvador wrote:

> On Fri, Jan 12, 2018 at 11:06 AM, Alex Stewart <alex.stewart at ni.com> wrote:
>> * Move the hwclock.sh initscript from the busybox recipe to util-linux.
>>    This script is generally useful for distros that get their hwclock
>>    implementation from sources other than busybox and we follow debian's
>>    example by providing it in util-linux.
>>
>> :busybox/*
>> * Remove the busybox-hwclock package, as it no longer has a purpose.
>> * If busybox is configured to include hwclock, the busybox package will
>>    RDEPEND on util-linux-hwclock-init.
>>
>> :util-linux/*
>> * Provide the hwclock.sh script in util-linux-hwclock-init, which can be
>>    pulled by any packages that depend on its functionality.
>> * util-linux-hwclock RDEPENDS on util-linux-hwclock-init for its
>>    initscript.
>>
>> Signed-off-by: Alex Stewart <alex.stewart at ni.com>
>> Acked-by: Haris Okanovic <haris.okanovic at ni.com>
>> Acked-by: Adrian Ratiu <adrian.ratiu at ni.com>
>> Acked-by: Ken Sharp <ken.sharp at ni.com>
>> Natinst-ReviewBoard-ID: 214983, 215755
> As mentioned by Christopher, it will increase the build time for
> everyone and there is real gain in this this change. If you want to
> disable the busybox dependency then move hwclock.sh to a hwclock-init
> package that has no standard dependencies and is allarch and make
> busybox and util-linux to depends on it.
>
Either solution works for my purposes, so I'm fine breaking hwclock.sh 
out as its own recipe. It seems degenerative to break such a small piece 
of common-code into its own recipe, but I suppose there isn't a better 
solution if util-linux isn't commonly built.

-- 
Alex Stewart
Software Engineer - LabVIEW Real-Time
National Instruments

alex.stewart at ni.com