On 01/08/2013 02:44 AM, Chris Larson wrote: > On Mon, Jan 7, 2013 at 2:36 PM, Florin Sarbu > > wrote: > > in alsa-utils_1.0.25.bb , I see this > comment: > > # We omit alsaconf, because > # a) this is a bash script > # b) it creates config files not suitable for OE-based distros > > Trying out on master, I get a do_rootfs error because bash is not > a rundep of alsa-utils, and there is nothing in the alsa-utils > recipe that actually does the above mentioned removal of the > alsaconf script. Looking in the git log, I only see that the > comment has suddenly appeared in an older alsa-utils recipe and > has been carried along since with every update to the recipe, and > not left over there as an artifact of some older commit as I expected. > Can anyone shed some light on this? > > > > From some quick git log -S commands in the old openembedded classic > repository, it seems to have originated > with 7653e980f66b4bd147df308f9b04d6b197382d6a from 2005, and the line > below the comment was 'FILES_${PN} = ""', so presumably that was what > left it out — none of the files patterns matched it. Okay, so it seems that the comment actually said at that time that the alsa-utils generated rpm won't contain alsaconf. And bellow, this bash script gets into the alsa-utils-alsaconf rpm. Unfortunately, even though there is this separation, the alsa-utils rpm depends on the alsa-utils-alsaconf rpm so it's pretty much the same thing as do_rootfs is concerned. (missing bash dependency). How would you guys see this resolved upstream? Add bash as a dep, remove alsa-utils-alsaconf as a dep or alsa-utils (as the comment says the script is not suitable for OE-based distros) or check the alsaconf script for bashisms and fix if any and make alsa-utils rely on a busybox shell interpreter if possible (with the modification of the !/bin/bash shabang from the alsaconf script to /bin/sh)? Florin > -- > Christopher Larson