[Dnsmasq-discuss] Announce: dnsmasq-2.27

[Simon Kelley]

Changelog below.

Cheers,

Simon.


            Tweaked DHCP behaviour when a client attempts to renew a
               	    lease which dnsmasq doesn't know about. Previously
that
            would always result in a DHCPNAK. Now, in dhcp-authoritative
            mode, the lease will be created, if it's legal. This makes
            dnsmasq work better if the lease database is lost, for
            example on an OpenWRT system which reboots. Thanks to
            Stephen Rose for work on this.

            Added the ability to support RFC-3442 style destination
            descriptors in dhcp-options. This makes classless static
            routes easy to do, eg dhcp-option=121,192.168.1.0/24,1.2.3.4

            Added error-checking to the code which writes the lease
            file. If this fails for any reason, an error is logged,
            and a retry occurs after one minute. This should improve
            things eg when a filesystem is full. Thanks to Jens Holze
            for the bug report.

            Fixed breakage of the "/#/ matches any domain" facility
            which happened in 2.24. Thanks to Peter Surda for the bug
            report.

            Use "size_t" and "ssize_t" types where appropriate in the
            code.

            Fix buggy CNAME handling in mixed IPv4 and IPv6
            queries. Thanks to Andreas Pelme for help finding that.

            Added some code to attempt to re-transmit DNS queries when
            a network interface comes up.  This helps on DoD links,
            where frequently the packet which triggers dialling is
            a DNS query, which then gets lost. By re-sending, we can
            avoid the lookup failing. This function is only active
            when netlink support is compiled in, and therefore only
            under Linux. Thanks to Jean Wolter for help with this.

            Tweaked the DHCP tag-matching code to work correctly with
            NOT-tag conditions. Thanks to Lutz Pressler for finding
            the bug.

            Generalised netid-tag matching in dhcp-range statements to
            allow more than one tag.

            Added --dhcp-mac to do MAC address matching in the same
            way as vendorclass and userclass matching. A good
            suggestion from Lutz Pressler.

            Add workaround for buggy early Microsoft DHCP clients
            which need zero-termination in string options.
            Thanks to Fabiano Pires for help with this.

            Generalised the DHCP code to cope with any hardware
            address type, at least on Linux. *BSD is still limited to
            ethernet only.