nvidia-drivers-* : src_install() { local MLTEST=$(type dyn_unpack) (...) if [[ "${MLTEST/set_abi}" == "${MLTEST}" ]] && has_multilib_profile ; then the dyn_unpack thing was added here: http://sources.gentoo.org/viewcvs.py/gentoo-x86/media-video/nvidia-glx/nvidia-glx-1.0.6629-r3.ebuild?hideattic=0&r1=1.4&r2=1.5 the MLTEST at all seems to first apear in this ebuild: http://sources.gentoo.org/viewcvs.py/gentoo-x86/media-video/nvidia-glx/nvidia-glx-1.0.6629-r2.ebuild?hideattic=0&rev=1.1&view=markup I /think/ this was added to check for some particular abi-related functionality in portage. Whatever it was seems to be outdated for quite some time as I didn't find "set_abi" in any dyn_unpack implementation since portage repo was moved to svn, but I didn't really check that carefully as it's wrong anyway. Please KILL.
Until someone from the CC list can explain what was intended by it, I'm not removing it. @eradicator: You added this code. Can you provide some details? @amd64: This code is for you guys... Can you explain?
IIRC, MLTEST was used in a portage branch maybe in 2003? 2004? when I was employed by GenUX to develop eselect-compiler, eselect-binutils, and add better multilib abstraction to portage. In short, GenUX pushed for things that were impractical, they went belly-up, and all that code entered bit-rot. I think it's safe just to - local MLTEST=$(type dyn_unpack) - if [[ "${MLTEST/set_abi}" == "${MLTEST}" ]] && has_multilib_profile ; + if has_multilib_profile ;
Unless the maintainers do not understand why this is a problem, or they are being uncooperative, there is no reason to add QA to this bug. Please don't add us back unless one of the above conditions is met. If we don't follow those guidelines, we might as well CC QA on every single bug in bugzilla. Thanks
I am no longer maintaining nvidia-*drivers in an way. Doug has been doing an excellent job and I didn't have much to do. Removing myself from CC.
fixed in 169.09-r1, the change will propagate to other versions as well.