># no /boot? try to mount.. >if [ ! -d /boot ] ; then O.k.- the directory doesn't exist, because the device isn't mounted? At least not on my system. ;) Also, there are unintended sideeffects when the ebuild gets unmerged, but /boot is/isn't mounted. May I suggest to install the stuff in /usr and let the user copy the binary to /boot via pkg_config()!?
same for sys-apps/memtest86 btw.
my suggest to check if /boot is mounted umountboot=0 ismount=$(mount | grep /boot) if [ -z "$ismount" ]; then #/boot isn't mounted umountboot=1 mount /boot >/dev/null fi #now /boot is mounted and we going on to install memtest in /boot, if /boot was already mounted this will skip mounting
with "/usr/portage/eclass/mount-boot.eclass" this problems seems to be solved. perhaps you can confirm that and close this bug, to clean the bug-database.
There's still a problem. The ebuild doesn't check, if there's enough space left on the boot partition.
memtest's use the mount-boot eclass so /boot is mounted properly as for not-enough-space ... that's a portage bug (if you really want to consider it a bug) which is handled elsewhere