Comment 3 for bug 204594

+ [ -n "$sig" -o "$sig" = 15 -o "$sig" = TERM ] && is_term_sig=yes

Nack on this change; I know it's standard practice for the LSB functions to explode miserably under set -e, but killproc currently does not do this, and init scripts may reasonably rely on a set-e-clean killproc. In fact, there's one init script on my machine, /etc/init.d/atd, which does - the function doesn't behave in an *LSB*-clean manner under set -e, but it does function, so this would be a regression.

Everything else changed here appears to be an improvement, so if you can correct this regression I'm ok with this.