On Mon, Jun 16, 2008 at 07:59:50AM -0400, James Carlson wrote: > Ceri Davies writes: > > I've been bitten this morning by bug 6296119 "lsvcrun doesn't seem to > > honor #! interpreter specifications" which was closed as "Not a Defect" > > some time ago. I'm about to beg to differ, although obviously I haven't > > seen why that was considered "not a defect". > > It's "not a defect" because a reference to the S9 implementation of > /sbin/rc2 showed that it did this: > > 76 for f in /etc/rc2.d/S*; do > 77 if [ -s $f ]; then > 78 case $f in > 79 *.sh) . $f ;; > 80 *) /sbin/sh $f start ;; > 81 esac > 82 fi > 83 done > > In other words, it simply has _never_ done what the user thought it > would do.
But it did do one thing that lsvcrun does not: it sourced files ending in .sh into the *same* shell process running /sbin/rc2. Years ago (long before SMF) I saw a service that had an S*.sh file that would set environment variables that automountd would then use. lsvcrun always starts a new shell process, which breaks that feature of S/K*.sh: such scripts now cannot affect the environment of subsequent S/K*.sh scripts. I'd think that /sbin/rc? should have been kept around, using svccfg(1M) or a utility program to update SMF when each [KS]* script is run. Nico --
