Skipping any file with "." in it seems a bit harsh. rc.subr is used by the scripts in /usr/local/etc/rc.d too, including anything users might write for local things. You're basically deciding that people can't have scripts named "mything.dev" "mything.test" "mything.prod" etc. without any way of knowing why this one case silently breaks. Seems not ideal to me.

as you can see from the function name, this is restricted to /etc/rc.d

/usr/local/etc/rc.d has other restrictions, which we only extended

however, we could log this skip case, so people know what's up

Ah, right. Still, might be better to only skip the things we know need to be skipped, IMHO.

I don't think we want this part, though. If you end up with .pkgnew or .pkgsave in the local scripts, then that's indicative of a real problem -- this is based on a vague recollection of a remark by @bapt that ports trying to mark anything as config file rather than using the sample mechanism should be smacked with an oversized broomstick.

We should also skip *.pkgnew here

we could move this to the part that has a *warning* about a bunch fragments:

run_rc_script()
…
        *[~#]|*.OLD|*.bak|*.orig|*,v)   # scratch file; skip          
                warn "Ignoring scratch file $_file"
                ;;

If you did this, add *~ to the list too... emacs backup files should be ignored too.