rpms/powertop/devel powertop-1.2-install-man-page.patch, NONE, 1.1 .cvsignore, 1.3, 1.4 powertop.spec, 1.2, 1.3 sources, 1.3, 1.4 powertop-1.1-build-fixes.patch, 1.1, NONE

[Matthias Clasen]

On Tue, 2007-05-15 at 19:15 +0200, Thorsten Leemhuis wrote:
> On 15.05.2007 19:04, Michael E Brown wrote:
> > On Tue, May 15, 2007 at 11:40:59AM -0400, Adam Jackson wrote:
> >> On Tue, 2007-05-15 at 17:14 +0200, Patrice Dumas wrote:
> >>> On Tue, May 15, 2007 at 09:01:41AM -0400, Adam Jackson wrote:
> >>>> Author: ajax
> >>>> +	mkdir -p ${DESTDIR}${MANDIR}
> >>>> +	cp powertop.1 ${DESTDIR}${MANDIR}
> >>> cp -p would keep timestamps. Maybe not worth it if powertop.1 is
> >>> generated, I haven't checked...
> >> It isn't.  Why would I care about timestamp?
> > Packaging guidelines state that it is preferrable to keep timestamps on
> > installed files the same as what was packaged.
> 
> Well, to give a better reasons than "because it's written": for multilib
> installs it's important that the timestamps are identical for files that
> are in both the i386 and x86_64 packages.
> 
> And (in the long term) making sure the timestamp didn't get changed
> might make things easier for presto as well.
> 

For once I have to agree with David Woodhouse: including timestamps into
file identity tests is seriously misguided.