The first paragraphs of the "1.e. Variables" section including code listing 5 are incorrect. RDEPEND="${DEPEND}" is mandatory, now. http://thread.gmane.org/gmane.linux.gentoo.devel/29509/
It's not GDP who maintain Developer Handbook, but Developer Relations. GDP cares only for doc/ content. Reassigning.
#turn off glob expansion from here on in to prevent *'s and ? in the DEPEND #syntax from getting expanded :) #check eclass rdepends also. set -f if [ "${RDEPEND-unset}" == "unset" ] && [ "${E_RDEPEND-unset}" == "unset" ] ; then export RDEPEND="${DEPEND} ${E_DEPEND}" debug-print "RDEPEND: not set... Setting to: ${DEPEND}" fi looks like it's in 2.1 to me...
Lukasz: Sorry, can't extract that from the enter_bug.cgi information. Alec: see e.g. bug 134555
(In reply to comment #0) > The first paragraphs of the "1.e. Variables" section including code listing 5 > are incorrect. RDEPEND="${DEPEND}" is mandatory, now. > > > http://thread.gmane.org/gmane.linux.gentoo.devel/29509/ > Guessing you are getting nailed by an eclass that sets E_RDEPEND, in which case the setting IS required, see related bugs about depend.apache.eclass
(In reply to comment #2) > looks like it's in 2.1 to me... The same code goes back at least as far as 2.0.53. Portage will set RDEPEND="${DEPEND}" automatically for an ebuild (not for an eclass) if RDEPEND is not set by the ebuild *and* not set by *any* inherited eclasses. This behavior should not be relied on and therefore ebuilds should set RDEPEND="${DEPEND}" explicity when necessary.
Please let us know if any more information is needed.
*** Bug 137629 has been marked as a duplicate of this bug. ***
current documentation is correct ... that section of documentation is talking about ebuilds, not eclasses
It says "Never set RDEPEND to DEPEND yourself in an ebuild" and that rule is incorrect.
removed the 'never' sentence
(In reply to comment #10) > removed the 'never' sentence > Not really enough as for example this is still there: # Not acceptable: RDEPEND="${DEPEND}" What Zac said in comment #5 should be added to the document.
Fixed in cvs. Patch reviewed by Zac.
except that portage is still fucked up and needs to be resolved on the mailing lists
(In reply to comment #13) > except that portage is still fucked up and needs to be resolved on the mailing > lists AFAIK, portage works fine when RDEPEND is set explicitly. Any change in implicit behavior changes the behavior of the entire tree and requires an EAPI bump.