Re: r5338 - in glibc-package/branches/eglibc-2.16/debian: . debhelper.in local rules.d
Le 26/10/2012 12:33, Adam Conrad a écrit :
> On Fri, Oct 26, 2012 at 08:35:26AM +0200, Aurelien Jarno wrote:
>> Le 26/10/2012 02:34, Adam Conrad a écrit :
>>> Author: adconrad
>>> Date: 2012-10-26 00:34:46 +0000 (Fri, 26 Oct 2012)
>>> New Revision: 5338
>>>
>>> Added:
>>> glibc-package/branches/eglibc-2.16/debian/debhelper.in/libc-bin.triggers
>>> glibc-package/branches/eglibc-2.16/debian/local/ldconfig_wrap
>>> Modified:
>>> glibc-package/branches/eglibc-2.16/debian/changelog
>>> glibc-package/branches/eglibc-2.16/debian/debhelper.in/libc-bin.lintian-overrides
>>> glibc-package/branches/eglibc-2.16/debian/debhelper.in/libc-bin.postinst
>>> glibc-package/branches/eglibc-2.16/debian/rules.d/debhelper.mk
>>> Log:
>>> Port over ldconfig trigger support from Ubuntu packages
>>
>> I don't think it's something needed. ldconfig has added sometimes ago an
>> auxilliary cache avoiding that repetitive call to ldconfig are
>> expensive. I think this should be reverted.
>
> When was this cache added? On a 2.15 system, this is definitely still a huge benefit:
According to #447609 it was in 2.7.
> root@shiva:~# export DPKG_MAINTSCRIPT_PACKAGE=foo
> root@shiva:~# time ldconfig
>
> real 0m0.051s
> user 0m0.008s
> sys 0m0.008s
> root@shiva:~# time ldconfig
>
> real 0m0.057s
> user 0m0.016s
> sys 0m0.008s
>
> root@shiva:~# unset DPKG_MAINTSCRIPT_PACKAGE
> root@shiva:~# time ldconfig
>
> real 0m0.756s
> user 0m0.023s
> sys 0m0.023s
> root@shiva:~# time ldconfig
>
> real 0m0.759s
> user 0m0.008s
> sys 0m0.031s
>
> This is on an ARM system that I haven't yet built 2.16 for (and these timings
> are significantly harder to reproduce on my IvyBridge laptop, obviously), but
> it certainly seems like the trigger is an obvious win here.
>
I recalled doing some tests (on amd64) and that the second call was
taking almost no time.
--
Aurelien Jarno GPG: 1024D/F1BCDB73
aurelien@aurel32.net http://www.aurel32.net
Reply to: