On Tue, 2006-03-21 at 09:44 -0700, Orion Poplawski wrote:
> Ralf Corsepius wrote:
> > On Mon, 2006-03-20 at 15:03 -0700, Orion Poplawski wrote:
> >> Ralf Corsepius wrote:
> >>> On Sun, 2006-03-19 at 18:40 -0500, Paul F. Johnson wrote:
>
> >> rm ${RPM_BUILD_ROOT}%{_infodir}/dir
> > If you think this is a clever idea - I prefer to call this unsafe and
> > error-prone.
>
> I think dropping the -f leads towards reproducible builds and less cruft
> in specs.
I tend to think the opposite :)
Whether the infodir/dir file is created in the buildroot, depends. So
it needs either testing for the file's existence before doing a rm
without -f or may need adding some BuildRequires that wouldn't be needed
for other purposes -> both result in _more_ cruft in the specfile. Not
worth it because that particular file is completely uninteresting
anyway.
Another somewhat offtopic example is when a file to be removed is read
only, causing rm to prompt and the build to hang. Yes, that should bite
the packager locally before the job gets stuck in the buildsys but we've
seen some cases of this happening there nevertheless.
--
fedora-extras-list mailing list
fedora-extras-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/fedora-extras-list
