Re: [PATCH 2/4] foomatic-filters: add a link to pass LSB printing test

[Richard Purdie <richard.purdie@linuxfoundation.org>]

On Fri, 2011-05-13 at 11:54 -0500, Mark Hatle wrote:
> On 5/13/11 11:49 AM, Saul Wold wrote:
> > On 05/13/2011 12:56 AM, Kang Kai wrote:
> >> From: Kang Kai<kai.kang@windriver.com>
> >>
> >> LSB printing test will test a foomatic-rip under cups' directory,
> >> so add it as link.
> >> Part fixes [Yocto 904]
> >>
> >> Signed-off-by: Kang Kai<kai.kang@windriver.com>
> >> ---
> >>   .../foomatic/foomatic-filters_4.0.7.bb             |    9 ++++++++-
> >>   1 files changed, 8 insertions(+), 1 deletions(-)
> >>
> >> diff --git a/meta/recipes-extended/foomatic/foomatic-filters_4.0.7.bb b/meta/recipes-extended/foomatic/foomatic-filters_4.0.7.bb
> >> index 8d70400..43e9c8d 100644
> >> --- a/meta/recipes-extended/foomatic/foomatic-filters_4.0.7.bb
> >> +++ b/meta/recipes-extended/foomatic/foomatic-filters_4.0.7.bb
> >> @@ -10,7 +10,7 @@ format using a printer-specific, but spooler-independent PPD file. \
> >>   "
> >>
> >>   DEPENDS += "cups perl libxml2"
> >> -PR = "r0"
> >> +PR = "r1"
> >>
> >>   LICENSE = "GPLv2+"
> >>   LIC_FILES_CHKSUM = "file://${WORKDIR}/foomatic-filters-${PV}/COPYING;md5=393a5ca445f6965873eca0259a17f833"
> >> @@ -20,4 +20,11 @@ SRC_URI = "http://www.openprinting.org/download/foomatic/foomatic-filters-${PV}.
> >>   SRC_URI[md5sum] = "20abd25c1c98b2dae68709062a132a7d"
> >>   SRC_URI[sha256sum] = "090313fae40b177f505d9c9b93d7a4d7188b6d5d18b6ae41ab24903ac983478d"
> >>
> >> +do_install_append_linuxstdbase() {
> >> +    install -d ${D}${libdir}/cups/filter
> >> +    ln -sf ../../../bin/foomatic-rip ${D}${libdir}/cups/filter
> > Kai Kang:
> > 
> > This should really be an absolute patch instead of a relative, there are 
> > distros that might have a different exec_prefix
> 
> We should likely audit our environments for these items.  I know at Wind River
> our policy is all links are relative, we do not allow absolute links as the
> values usually cause problems in sysroots, or external SDK environments.
> 
> (Note, pseudo has a mechanism internal to it that ensures that absolute links
> don't point off to system files.. but we still avoid it.)

If we want to use relative paths, they need to be calculated using the
current layout. I can think of at least one userbase which is going to
want to set all the *prefix variables to "/" instead of "/usr" and I'd
prefer not to introduce additional problems for them.

Cheers,

Richard