Log message:
s/ucspi-tcp/ucspi-tcp6/g in dependencies. Bump PKGREVISIONs.

Log message:
Catch up to qmail-run:

- Collapse redundant code for invoking service-specific rc.d scripts.
- Don't try to run a service's rc.d script if it isn't enabled in rc.conf.
- Prefix "nb" to procnames.

Bump version.

Log message:
Catch up to qmail-run:

- Wrap long command lines for readability
- Document where we set procname=${name} for rc.d
- Detach long-running processes from controlling terminal
- Configurable path to tcpserver
- Configurable user and group names:

DJBDNS_AXFR_USER?=	axfrdns
DJBDNS_CACHE_USER?=	dnscache
DJBDNS_RBL_USER?=	rbldns
DJBDNS_TINY_USER?=	tinydns
DJBDNS_DJBDNS_GROUP?=	djbdns

Bump version.

Log message:
Remove mk/find-prefix.mk usage from the net category.

The find-prefix infrastructure was required in a pkgviews world where
packages installed from pkgsrc could have different installation
prefixes, and this was a way for a dependency prefix to be determined.

Now that pkgviews has been removed there is no longer any need for the
overhead of this infrastructure.  Instead we use BUILDLINK_PREFIX.pkg
for dependencies pulled in via buildlink, or LOCALBASE/PREFIX where the
dependency is coming from pkgsrc.

Provides a reasonable performance win due to the reduction of `pkg_info
-qp` calls, some of which were redundant anyway as they were duplicating
the same information provided by BUILDLINK_PREFIX.pkg.

Log message:
Based on feedback from joerg@ and riastradh@, don't save a persistent
seed file on first startup after all. Primary rationale: it is not
desirable that the predictability of the first boot (whatever that
may happen to be) force an upper bound on the unpredictability of
random seeds for all future dnscache instances. Instead, as before,
let dnscache read 128 new bytes on every startup. Just give it
/dev/urandom instead of /dev/random.

While here, abolish references to LOCALBASE. Bump version.

Log message:
On rebooting a Xen VPS, dnscache said it started, but failed to
respond to queries and wrote nothing to the logs. This would have
been more fun to debug if my production system hadn't been relying
on it for name resolution. The problem was a blocking read from
/dev/random.

I'm the one who put that there, over 10 years ago. Do we need it?

From my reading of <http://cr.yp.to/djbdns/dnscache-conf.html> and
dnscache-conf.c, the seed needn't consist of top-notch randomness.
A sysadmin following DJB to the letter (rather than using this
djbdns-run package) would run dnscache-conf once to configure a
dnscache instance. dnscache-conf would cook 128 bytes of randomness
-- without direct assistance from any kernel device -- and write
it to a file. dnscache would read the same file on every startup.

Therefore, we...

1. Generate a random seed iff it doesn't exist (not on every startup)
2. Generate it from /dev/urandom instead of /dev/random

...so that the configuration we generate more nearly matches that of
DJB's documented procedure, and the kind of problem I encountered
is much more difficult to trigger.

While here, add LICENSE (2-clause-bsd) and otherwise placate pkglint.

Bump version.

Log message:
Remove pkgviews: don't set PKG_INSTALLATION_TYPES in Makefiles.

Log message:
Fix the boot-time error (which also resulted in ntpdate failing):

rcorder: Circular dependency on provision `named' in file `/etc/rc.d/ntpdate'.

Fix it by aping -r1.22 of src/etc/rc.d/named, which had replaced
"SERVERS" with "NETWORKING mountcritremote syslogd". Problem and
fix reported by Nathan Arthur.

Bump version by exactly nine years.

Log message:
Remove example rc.d scripts from PLISTs.

These are now handled dynamically if INIT_SYSTEM is set to "rc.d", or
ignored otherwise.

Log message:
Drop superfluous PKG_DESTDIR_SUPPORT, "user-destdir" is default these days.