Summary of changes since 1.2.0:
* dns/rdtypes/ANY/TXT.py (TXT.from_text): The masterfile parser
incorrectly rejected TXT records where a value was not quoted.
* dns/message.py: Added make_response(), which creates a skeletal
response for the specified query. Added opcode() and set_opcode()
convenience methods to the Message class. Added the request_payload
attribute to the Message class.
* dns/zone.py (from_xfr): dns.zone.from_xfr() in relativization
mode incorrectly set zone.origin to the empty name.
* dns/name.py (Name.to_wire): The 'file' parameter to
Name.to_wire() is now optional; if omitted, the wire form will
be returned as the value of the function.
* dns/message.py (Message.find_rrset): find_rrset() now uses an
index, vastly improving the from_wire() performance of large
messages such as zone transfers.
* dns/query.py: sending queries to a nameserver via IPv6 now
works.
* dns/inet.py (af_for_address): Add af_for_address(), which looks
at a textual-form address and attempts to determine which address
family it is.
* dns/query.py: the default for the 'af' parameter of the udp(),
tcp(), and xfr() functions has been changed from AF_INET to None,
which causes dns.inet.af_for_address() to be used to determine the
address family. If dns.inet.af_for_address() can't figure it out,
we fall back to AF_INET and hope for the best.
* dns/rdtypes/ANY/NSEC.py (NSEC.from_text): The NSEC text format
does not allow specifying types by number, so we shouldn't either.
* dns/renderer.py: the renderer module didn't import random,
causing an exception to be raised if a query id wasn't provided
when a Renderer was created.
* dns/resolver.py (Resolver.query): the resolver wasn't catching
dns.exception.Timeout, so a timeout erroneously caused the whole
resolution to fail instead of just going on to the next server.
* dns/rdtypes/ANY/LOC.py (LOC.from_text): LOC milliseconds values
were converted incorrectly if the length of the milliseconds
string was less than 3.
* dns/update.py (Update.delete): We erroneously specified a
"deleting" value of dns.rdatatype.NONE instead of
dns.rdataclass.NONE when the thing being deleted was either an
Rdataset instance or an Rdata instance.
* dns/rdtypes/ANY/SSHFP.py: Added support for the proposed SSHFP
RR type.
* dns/rdata.py (from_text): The masterfile reader did not
accept the unknown RR syntax when used with a known RR type.
* dns/name.py (from_text): dns.name.from_text() did not raise
an exception if a backslash escape ended prematurely.
* dns/zone.py (_MasterReader._rr_line): The masterfile reader
erroneously treated lines starting with leading whitespace but
not having any RR definition as an error. It now treats
them like a blank line (which is not an error).
* Added support for new DNSSEC types RRSIG, NSEC, and DNSKEY.
* dns/query.py (_connect): Windows returns EWOULDBLOCK instead
of EINPROGRESS when trying to connect a nonblocking socket.
* dns/rdtypes/ANY/LOC.py (LOC.to_wire): We encoded and decoded LOC
incorrectly, since we were interpreting the values of altitiude,
size, hprec, and vprec in meters instead of centimeters.
* dns/rdtypes/IN/WKS.py (WKS.from_wire): The WKS protocol value is
encoded with just one octet, not two!
* dns/resolver.py (Cache.maybe_clean): The cleaner deleted items
from the dictionary while iterating it, causing a RuntimeError
to be raised. Thanks to Mark R. Levinson for the bug report,
regression test, and fix.
Changes in 1.2.0:
- Timeout support has been overhauled. It now works on Python 2.2 as
well as 2.3, and the total time spent processing a query is now much
more controllable than before since the timeout is now on the whole
query instead of just on individual network calls.
- The master file reader now gives the filename and line number of
the offending input when a syntax error occurs.
- $INCLUDE is now supported in DNS master files. Processing of
$INCLUDE can also be disabled if desired.
- BIND 8 style TTLs, e.g. "1w2d3h4m5s", are accepted when reading a
master file, but will never be emitted.
- Basic zone sanity checks are made after a zone is
loaded. Specifically, the zone must have SOA and NS rdatasets at its
origin.
Inspired by FreeBSD "ports".
Fix the PLISTs accordingly.
Also, while at it, remove now obsolete compileall.py calls in post-install
targets and insure that extension.mk is in included before builinlinks of
other Python modules.
Discussed with/ok'ed by drochner@.
Summary of changes from 1.0.0:
* Message sections are now lists of RRsets, not lists of nodes.
* Nodes no longer have names; owner names are associated with
nodes in the Zone object's nodes dictionary.
* Many tests have been added to the test suite; dnspython 1.0.0
had 47 tests, 1.1.0 has 275. The improved testing uncovered a
number of bugs, all of which have been fixed.
* The NameDict class provides a dictionary whose keys are DNS
names. In addition to behaving like a normal Python dictionary,
it also provides the get_deepest_match() method. If, for
example, you had a dictionary containing the keys foo.com and
com, then get_deepest_match() of the name a.b.foo.com would
match the foo.com key.
* A new Renderer class for those applications which want finer
control over the DNS wire format message generation process.
* Support for a "TooBig" exception if the size of wire format
output exceeds a specified limit.
* Zones now have find_rrset() and find_rdataset() convenience
methods. They let you retrieve rdata with the specified name
and type in one call, e.g.:
rrset = zone.find_rrset('foo', 'mx')
* Other new zone convenience methods include: find_node(),
delete_node(), delete_rdataset(), replace_rdataset(),
iterate_rdatasets(), and iterate_rdatas().
* get_ variants of find_ methods are provided; the difference is
that get_ methods return None if the desired object doesn't
exist, whereas the find_ methods raise an exception.
* Zones now have a to_file() method.
* The message and zone from_file() methods allow Unicode filenames
on platforms (and versions of python) which support
them. Universal newline support is also used if available.
* The Zone class now implements more of the standard mapping
interface. E.g. you can say zone.keys(), zone.get('name'),
zone.iteritems(), etc. __iter__() has been changed to iterate
the keys rather than values to match the standard mapping
interface's behavior.
* Rdatasets support more set operations
* Zone and Node factories may be specified, allowing applications
to subclass Zone or Node and yet still use the algorithms which
build zones from master files or AXFR data.
* dns.ipv6.inet_ntoa() now minimizes the text representation of
IPv6 addresses in the usual way,
e.g. "0000:0000:0000:0000:0000:0000:0000:0001" is minimized to
"::1".
* dns.query functions now take an optional address family parameter.
All known bugs from 1.0.0 are fixed in this release.
Changes from 1.0.0b3:
* dns/rdata.py: Rdatas now implement rich comparisons instead of
__cmp__.
* dns/name.py: Names now implement rich comparisons instead of
__cmp__.
* dns/inet.py (inet_ntop): Always use our code, since the code
in the socket module doesn't support AF_INET6 conversions if
IPv6 sockets are not available on the system.
* dns/resolver.py (Answer.__init__): A dangling CNAME chain was
not raising NoAnswer.
* Added a simple resolver Cache class.
* Added an expiration attribute to answer instances.
dnspython is a DNS toolkit for Python. It provides both high and low
level access to DNS. The high level classes perform queries for data
of a given name, type, and class, and return an answer set. The low
level classes allow direct manipulation of DNS zones, messages, names,
and records.
