On 10/03/12 01:21, Dave Reisner wrote: > On Fri, Mar 09, 2012 at 05:59:07PM +1000, Allan McRae wrote: >> Signed-off-by: Allan McRae <al...@archlinux.org> >> --- >> scripts/makepkg.sh.in | 7 +++++-- >> 1 file changed, 5 insertions(+), 2 deletions(-) >> >> diff --git a/scripts/makepkg.sh.in b/scripts/makepkg.sh.in >> index 8dd2d39..9cd8af8 100644 >> --- a/scripts/makepkg.sh.in >> +++ b/scripts/makepkg.sh.in >> @@ -2052,8 +2052,11 @@ readonly ALL_OFF BOLD BLUE GREEN RED YELLOW >> BUILDDIR=${_BUILDDIR:-$BUILDDIR} >> BUILDDIR=${BUILDDIR:-$startdir} #default to $startdir if undefined >> if [[ ! -d $BUILDDIR ]]; then >> - mkdir -p "$BUILDDIR" || >> - error "$(gettext "You do not have write permission to >> create packages in %s.")" "$BUILDDIR" >> + if ! mkdir -p "$BUILDDIR"; then >> + error "$(gettext "You do not have write permission to create >> packages in %s.")" "$BUILDDIR" > > Hrmmm, permissions is only one possible reason for mkdir to fail. the > manpage for the syscall and library wrapper list a bunch more. Since > mkdir will happily report the error for us (with localization), should > we just ditch our own possibly-inaccurate error messaging? >
I guess mkdir is one of the functions we want a safe wrapper for and we could have a generic "Failed to create directory: %s" type message there. So I will just fix the bug for now under the assumption a bigger fix is coming... Allan