* Licensed under GPLv3 or later.

* mixal

** Location of listing file

By default, mixal creates listing file in the current working directory.

* mixsim

** The .mixsim file
When started in terminal mode, the utility searches for file `.mixsim'
in the current working directory and reads it, if present. The file
may contain any mixsim commands.

** The set of terminal commands is improved to make it compatible with GDB.

** New command, PASSCOUNT
Set the number of crossings after whcih to clear the breakpoint.

** Removed optional second argument to BREAK.
The functionality of `BREAK <ADDR> <COUNT>' is now obtained by two
commands:

BREAK <ADDR>
PASSCOUNT <N> <COUNT>

** New command DELETE.
An alias to CLEAR.

** New command IGNORE
Sets the number of hits to ignore before enabling the breakpoint.

** New commands ENABLE and DISABLE
Enable or disable the breakpoint with the given number.

** New command RUN
It is an alias for GO.

** New command SOURCE
Execute a file of mixsim commands.

** New command ADDRESS.
This command takes as its argument a breakpoint manipulation command
(DELETE, IGNORE, ENABLE, DISABLE, PASSCOUNT) with arguments
appropriate for the associated command.  However, the first argument
of the corresponding command is treated as MIX address, instead of a
breakpoint number.

** New command INFO.
An alias to LIST.

** New command SHELL.
Executes a subordinate shell or a shell command. Abbreviated to !.

** NEXT and STEP take optional argument.
The argument, if given, specifies number of instructions to
execute. Default is 1.

** INFO BREAK takes an optional argument specifying breakpoint number to list.
If prefixed with ADDRESS, it treats the argument as a MIX location and
lists all breakpoints set on that location.
	  

* Configuration file raddb/config

The syslog statement takes an optional 4th argument specifying syslog
tag to use, e.g.:

    channel default {
	    syslog local1.info radiusd;
    };

* New attributes

** GNU-Server-Address

Holds IP address of the RADIUS server that recieved the
request. Notice, that the value of this attribute is "0.0.0.0" if
there are no `listen' statement in your `raddb/config'.

** GNU-Server-Port

Holds UDP port number of the RADIUS server that recieved the request.

* Automake function AM_GNU_RADIUS is provided, for checking if
GNU Radius is installed from configure.ac scripts.

* Guile support requires Gule version 1.8 or later.

* Bugfixes
** Pass NAS-IP-Address to mlc_stop_query
	  

Initial release.
	  

* Use IPv6 URLs as proposed by RFC 2732

The address part of an IPv6 URL must be either an IPv6 address in
numeric notation *enclosed in square brackets* or a host name, e.g.:

  inet6://[::1]:1100
	  

First actual release.
	  

* Support SNMPv3
* Default SNMP version is 2c
* Use newer Grecs
* Drop unneeded dependencies
* Variable assignments in expressions
* Detection of SNMP counter overflows
* Comma operator
* Support for indexed MIBs

This feature allows you to use symbolic names instead of the fixed MIBs
for MIBs that are part of SNMP subtrees.  For example, to get number of
packets sent over eth0 into variable "out", you would do the following:

    table iftable IF-MIB::ifDescr;
    variable out "IF-MIB::ifOutUcastPkts.$iftable[eth1]";

The first statement converts the subtree into a "table" named "iftable".
The second statement references an entry in this table that has
the value "eth1".  For example, if the SNMP tree has the following MIB

    IF-MIB::ifDescr.10: eth0

then the expression "$iftable[eth]" yields "10"

* Assertion syntax changed.

The assertion statement takes a single argument, which must be a
string consisting of the following three parts:

   <oid: string> [!]<opcode>[/i] <value: string>

The <opcode> part can be either an arithmetical operator (=, <, <=, >,
>=), or any of the following string operators:

    eq          string equality
    ne          string inequality
    prefix      oid value must begin with <value>
    suffix      oid value must end with <value>
    glob        <value> is a glob(7) pattern that oid value must match

Each of these can be suffixed with "/i" to request case-insensitive comparison.

A "!" in front of opcode reverts its meaning.

The <value> part must not include the type prefix.
	  

* The program is installed in /usr/bin

Prior versions went to /usr/sbin, you will need to remove them
manually.

* Print /etc/issue before the prompt
* Sleep after incorrect password is input
* New options: -i (--issue), -s (--sleep), -c (--clear)
* Add a manpage
* Improve error checking and reporting
	  

* Fix threshold calculation in the cleaner module.
* Refuse to start if another copy is already running.
	  

* ftp

An old inability to allow other names than the canonical name has
been corrected.  This means that a machine entry in the .netrc file
will now be used as expected.  Previously any alias name was replaced
by the corresponding canonical name, before reading the .netrc file.

The internal command `hash' accepts a suffixed letter to the size
argument, like `12k', instead of 12288.  Made a minor change to the
syntax of the command itself, allowing size changes independently
of activation of hash markings.  After a transfer the summary gives
the speed as `Mbytes/s', `kbytes/s', or `bytes/s'.

The .netrc file can be overridden by the environment variable NETRC.
Of even higher precedence is the new option `-N/--netrc'.  The access
to the resulting file, whatever method, is now denied unless it is a
regular file.

* ifconfig

Better command line parsing on BSD and Solaris systems.  Touch only
changeable flags on all systems.

* logger

The ability to use numerical facilities is restored to full range.

* ping, ping6

The ability to specify a pattern as payload is corrected.

* syslogd

A new switch `-T/--local-time' makes the service ignore a time
stamp passed on by the remote host, recording instead the local
time at the moment the message was received.  As a short form of
`--pidfile', the switch `-P' is new.

In common with other syslogd implementations, rsyslogd and sysklogd,
there has for a long time existed an attack vector based on large
facility numbers, made public in CVE-2014-3684.  This is now mended
in our code base.

* telnetd

The ability to autologin a client, without using authentication,
is now functional in the expected manner, i.e., the prompt for a
user name is suppressed in favour of an immediate password prompt.

In a setting where the client is using a UTF-8 encoding, it was
common to observe strange characters in most responses.  This was
caused by the server daemon, due to incomplete purging of internal
protocol data.  The issue should now be resolved.

* whois

Improved cooperation with servers like `whois.arin.net', `whois.eu',
and `whois.ripe.net'.
	  

* Convert textual frames to/from the currently used character set

The character set is deduced from the locale settings.  It can also be
set explicitly using the --charset option.

* New option --broken-8bit-encoding

Use this option for files where textual frames are stored as
ISO-8859-1, but are actually using another 8-bit encoding.  The
argument to this option is the name of the enchoding actually used.

In query mode, this option helps display such frames properly.

Use it with the --fixup option to fix such frames in the file.

* New option --encoding

Specifies encoding to store textual frames in ID3 tags.

* New option --fixup

Modifies the ID3 v2 tag so that it can be understood by most devices.

* The --convert option can be used to remove unnecessary ID3 formats.

For example, if the file input.mp3 contains both version 1 and 2 tags,
the following will remove version 1 tags:

  idest --convert=2 input.mp3

* Fix processing of unknown frames.

* Fix operation of setpic and pic modules with Guile 2.x.

* Improve documentation
	  

* Implement heartbeat event

* Add auxiliary program: ifactive

* Fix file descriptor leak
	  

Convert byte-like objects to UTF-8 strings.
	  

Rewrite as a stand-alone snmpd agent.
	  

Initial release.
	  

Initial release.
	  

Initial release.
	  

Initial release.
	  

* Minor fixes in mysqlstat-setup
* Provide default values for NULL columns
	  

* Drop support for Varnish versions prior to 6.0.0
	  

Fix handling of archivation requests.
	  

* New command line option --done (-d)

This option is a counterpart of --init and supplies a cleanup
expression, i.e. an expression that will be evaluated when the
main loop has iterated over all nodes in the tree.

* New parser: DHCPD

A parser for dhcpd.conf file.

* Drop support for Guile versions prior to 2.2.0
	  

* Change bugtracker address.
	  

	- Change bugtracker address.
	  

* Fix parsing of po header fields

* Use the Language header instead of Language-Team.

Support for Language-Team and language tables is retained for
backward compatibility.

* Organizational change: repository moved to git

The repository along with links for cloning can be viewed online at
http://git.gnu.org.ua/cgit/wyslij-po.git.

See https://puszcza.gnu.org.ua/git/?group=wyslij-po for instructions.
	  

	- Change bugtracker address.
	  

	- Change bugtracker address.
	  

	- Change bugtracker address.
	  

	- Change bugtracker address.
	  

	- Change bugtracker address
	  

	- Change bugtracker address
	  

* New configuration statement: sentinel

The "sentinel" statement declares the name of a file which, when
present in the destination directory, informs vcsync that this
directory is exempt from the normal synchronization process.  The
presense of the sentinel file is an indication that the destination
directory is updated by other means.

* Full documentation added
	  

* Use explicit DESTROY with AUTOLOADs
	  

* Bugfixes in the gcide module

* Fix searches in the gcider utility

* Support for Python 2 has been withdrawn
	  

* Syslog support

New option '-S FACILITY' switches diagnostic output to the given
syslog facility.
	  

* Support for Guile 2.2

* Bugfixes

* Improved testsuite
	  

	- Fix operation without -d option
	  

* Multiple start functions are allowed

The '--main' option can be given multiple times.  A separate graph
will be drawn for each function given as its argument.

* New option --target=FUNCTION

If this option is given, the produced graph will contain only paths
leading from start function (or functions) to the given FUNCTION.

Multiple '--target' options are allowed.

* New output format: dot

The '-f dot' (or '--format=dot') option instructs cflow to output
graph as a description in DOT language, suitable as input to graphviz
programs.

* cflow-mode: new commands for navigating in the graph:

  c   go to the calling function
  n   go to the next function at the same nesting level
  p   go to the previous function at the same nesting level

* Bugfixes:

** CVE-2019-16165
** CVE-2019-16166
** Fix parsing of K&R style function declarations
** Improve parsing of typecasts
** Fix recursive call detection
	  

* Hostgroup stack

Rex now maintains a stack of hostgroups.  Once you push a hostgroup
name on stack, that hostgroup will be used by any subsequent rex
command, unless it is given the -g option.  The new command "group"
(see below) provides ways to pop items off the stack, swap arbitrary
element with the top of stack, select new hostgroups on the fly, etc.

* New command: rex group

The "rex group" command has the following forms:

  rex group push GROUP
    Push GROUP on stack.

  rex group pop
    Pop the topmost group off the stack.

  rex group swap N
    Exchange top of the stack with the Nth element (0-based).

  rex group drop N
    Remove Nth element from the stack.

  rex group select COMMAND ARGS...
    Create on top of the stack a temporary group that contains
    those hosts from the current hostgroup where the supplied
    shell command returns success, i.e. exits with code 0.

  rex group show
    List the contents of the hostgroup stack.

* Add configuration function for overriding PTR records from rc files
	  

* Support for Slackware 15.0
	  

* pam_fshadow: skip-password option

Based on the proposal of Mirsad Goran Todorovac, the new option
skip-password instructs pam_fshadow to check whether the user
being authenticated is present in the passwd and/or shadow files,
without verifying his password.  This way pam_fshadow can be used as
an auxiliary module in the stack, actual authentication being
performed by one of the modules before it.
	  

	- Handle the 'resolvers' section.
	- New class methods for declaring (and undeclaring) sections.
	- Fix changing the argv of a Node.
	  

* Support for Varnish 7.1
	  

* New configuration keywords

** sigterm SIG

Available for use in "component" sections.  This statement defines
signal which pies should send to the running component instance in
order to terminate it.  Defaults to SIGTERM.

* Fix the component shutdown sequence

The shutdown sequence is determined taking into account dependencies
between components, so that all dependent components are stopped
before their prerequisite components.

* Fallback log file

Fallback log file is a place where pies writes out of band log messages,
i.e. messages about not being able to open syslog socket or send logs
to it.  Regular log messages are diverted to this file if syslog was
requested, but cannot be used because of a permanent error.

* Bugfixes

** Fix piesctl config reload

** Fix configuration preprocessing.

** Varios fixes in REST API server.
	  

* Support for Varnish 7.1
	  

* Support for Varnish 7.1
	  

* Support for Varnish 7.1
* Drop support for Varnish versions prior to 6.0.0
	  

* Support for Varnish 7.1
	  

* Support for Varnish 7.1
	  

* Support for Varnish 7.1
* Drop support for Varnish versions prior to 6.0.0.
	  

* Support for Varnish 7.1
	  

* Fix https://puszcza.gnu.org.ua/bugs/?562
	  

Fix deadlock.
	  

* Fix handing of the Connection: HTTP handler (both server and client)
	  

* Diagnostic directives: $$warning and $$error

The $$warning directive emits warning message.  It does not alter exit
status in any way.

The $$error directive reports a fatal error and sets exit status to
65 (or any other, if supplied as argument).

After both directives, processing is resumed at the next line.

* $$exit

New directive $$exit causes immediate termination of the program.
Decimal exit code may be supplied as argument.

* New directive: $$eval

The text between $$eval and $$end is expanded and the resulting
expansion is scanned again, producing the actual output.  This makes
it possible to create variable names on the fly and obtain their values.
Useful in loops, e.g.:

  $$loop I 0 1 2 3 4 5 6 7
  $$  eval
  \$\$ ifset VAR_$I
  Expand \$VAR_$I;
  \$\$ endif
  $$ end
  $$end

* Bugfixes

** Fix closing the $$range loop.
	  

Bootstrapped with mailfromd 8.16.93. This fixes configure failure
on FreeBSD.
	  

Bootstrapped with mailfromd 8.16.93. This fixes configure failure
on FreeBSD.
	  

* Fail when building GNU tar, if the platform supports 64-bit time_t
  but the build uses only 32-bit time_t.

* Leave the devmajor and devminor fields empty (rather than zero) for
  non-special files, as this is more compatible with traditional tar.

* Bug fixes

** Fix interaction of --update with --wildcards.

** When extracting archives into an empty directory, do not create
   hard links to files outside that directory.

** Handle partial reads from regular files.

** Warn "file changed as we read it" less often.
   Formerly, tar warned if the file's size or ctime changed.
   However, this generated a false positive if tar read a file
   while another process hard-linked to it, changing its ctime.
   Now, tar warns if the file's size, mtime, user ID, group ID,
   or mode changes.  Although neither heuristic is perfect,
   the new one should work better in practice.

** Fix --ignore-failed-read to ignore file-changed read errors
   as far as exit status is concerned.  You can now suppress file-changed
   issues entirely with --ignore-failed-read --warning=no-file-changed.

** Fix --remove-files to not remove a file that changed while we read it.

** Fix --atime-preserve=replace to not fail if there was no need to replace,
   either because we did not read the file, or the atime did not change.

** Fix race when creating a parent directory while another process is
   also doing so.

** Fix handling of prefix keywords not followed by "." in pax headers.

** Fix handling of out-of-range sparse entries in pax headers.

** Fix handling of --transform='s/s/@/2'.

** Fix treatment of options ending in / in files-from list.

** Fix crash on 'tar --checkpoint-action exec=\"'.

** Fix low-memory crash when reading incremental dumps.

** Fix --exclude-vcs-ignores memory allocation misuse.
	  

* Fix compilation with gcc 12.2.0
	  

* Don't bail out if mailq or mailstats fails.
	  

* Fix check for hostname presence in the message
	  

* Compatibility .mf suffix not supported any more

* Begin and end handlers must be defined using prog keyword

* Module and include search paths

Since version 8.15, if a module was not found in module search path,
the search was retried using include search path.  This is no longer
the case: the two paths serve different purposes and don't interact
in any way.  MFL modules are searched in module search path only.

* Use of Sendmail macros in run mode

Sendmail macros can be defined in run mode by placing assignments
macro=value before the script file name, e.g.:

  mailfromd --run i=123 client_addr=::1 test.mfl foo bar baz

This example defines Sendmail macro "i" to "123" and "client_addr" to
"::1".  The words "foo", "bar", and "baz" will be passed to the
main function as positional parameters.

* Full IPv6 support

This causes changes in the following MFL functions (see also "New functions",
below):

** dns_query

 number dns_query (number TYPE, string DOMAIN; number SORT, number RESOLVE)

The semantics and possible values of the RESOLVE argument have
changed.  It used to be a boolean value.  Now its allowed values (as
defined in status.mfl module) are:

     'RESOLVE_NONE'
          Don't resolve hostnames to IP addresses.  This is the default.

     'RESOLVE_DFL'
          Resolve hostnames to IP addresses according to the address
          family of the SMTP session.  That is, use 'A' records if the
          client connected using the INET family (i.e.  connected to the
          IPv4 address), and use 'AAAA' records if the client connected
          to the IPv6 address.

     'RESOLVE_IP4'
          Resolve hostnames to IPv4 addresses ('A' records).

     'RESOLVE_IP6'
          Resolve hostnames to IPv6 addresses ('AAAA' records).

Values of these constants has been selected so that any existing
code using 0/1 as the value of this argument will work without
changes.

** primitive_resolve and resolve functions

 string primitive_resolve (string HOST; string DOMAIN, number FAMILY)
 string resolve (string HOST; string DOMAIN, number FAMILY)

The use of DOMAIN argument is deprecated.

By default the function selects the type of RR set to search for
using the address family of the SMTP connection: A is used for INET
(IPv4 addresses) and AAAA is used for INET6 (IPv6 addresses).

New argument FAMILY can be used to request particular RR type.  Its
possible values are:

     'RESOLVE_DFL'
          Look for 'A' or 'AAAA', depending on the connection type.
          This is the default.

     'RESOLVE_IP4'
          Resolve to IPv4 addresses ('A' records).

     'RESOLVE_IP6'
          Resolve to IPv6 addresses ('AAAA' records).

* New functions

** string is_ipstr(string S)

Returns 1 if S is a string representation of an IP address (IPv4 or
IPv6).

** string is_ip4str(string S)

Returns 1 if S is a string representation of an IPv4 address.

** string is_ip6str(string S)

Returns 1 if S is a string representation of an IPv6 address.

** string reverse_ipstr(string IP)

Returns a reversed representation of the IP address, suitable for
use in DNS labels.

** string tr(string SUBJ, string SET1, string SET2)

Transliterates characters in SUBJ by replacing all occurrences of the
characters found in SET1 with the positionally corresponding characters
in SET2.  Character ranges ('a-z') are allowed in both sets.  A character
range in SET1 translates to a corresponding character from the range in
SET2, e.g. tr(x, 'a-z', 'A-Z') translates string x to uppercase.

Character classes ([a-z], [[:alpha:]], etc) are allowed in SET1 and
translate to the corresponding single character from SET2, e.g.
tr(x, '[[:digit:]]', '_') replaces all decimal digits in string x with
underscores.

** string dc(string SUBJ, string SET)

Deletes from SUBJ characters that appear in SET.  The syntax of SET is as
described for SET1 above.

** string sq(string SUBJ, string SET)

"Squeeze repeats".  Replaces each sequence of a repeated character
that is listed in SET, with a single occurrence of that character.

* Changes in match_cidr function.

The match_cidr function is re-implemented as a built-in.

The module match_cidr.mfl is deprecated.

* Deprecated MFL modules

The following MFL modules are deprecated.  They are retained for
backward compatibility.  Existing code using any of these will compile
and work as in previous releases, except that a warning will be
printed to draw your attention to the fact.  You are advised to remove
any uses of these modules, as they will be removed in future versions:

** match_cidr.mfl

This module is not needed any more.

** is_ip.mfl

This module defines function is_ip() which is superseded by is_ipstr()
built-in (see above).

** revip.mfl

This module defines function revip() which is superseded by
reverse_ipstr() built-in.
	  

** Fix compilation with GDBM 1.18.1

** Support for Guile version 2.2.0 and later

Support for prior versions has been withdrawn.

** anubisusr requires GnuTLS

** New configuration statement: use-pam

Used in CONTROL section, this boolean statement enables or disables
the use of the Pluggable Authentication Module interface for
accounting and session management.

** New configuration statement: identd-keyfile

Sets the name of the file with shared keys used for decryting replies
from the auth service.  It is used in traditional mode if anubis
receives an encrypted response from the client's identd server
(e.g. if they are running pidentd with encryption).

** Bugfixes
	  

* Use of TLS in pop3d and imap4d

If not explicitly specified, the TLS mode to use (ondemand, connect,
etc.) is derived from the configured port.  E.g., for imap4d, port 143
implies ondemand mode, and port 993 implies connection mode.

The global tls-mode setting is used only when the mode cannot be
detemined otherwise, i.e. neither per-server tls-mode is given nor
the port gives any clues as to the TLS mode to use.

* Bugfixes

** movemail: fix handling of -P option.
** pop3d, imap4d: fix global timeout and transcript configuration statements
** pop3d, imap4d: global tls-mode is used only when the mode to use cannot
   be determined otherwise.
** pop3d, imap4d: if not set explicitly, derive tls mode from the port used.
** Improve file safety checking routine.
** Fix compilation with new gcc versions.
	  

* Micron is a part of GNU project

* Take into account DST changes when computing job startup time

* Minor fix in crontab parser
	  

* Automatic release type

Release type can be determined automatically, from version numbers of
the files to be uploaded.  To do so, version number semantics must
be declared in the configuration file using the "version_semantics"
statements:

  version_semantics gnu
     Assumes GNU version semantics: two-part version numbers
     (e.g. 1.0) or three-part numbers whose third value (patchlevel)
     is numerically less than 90 mean "stable" releases.  Three-part
     numbers with patchlevel greater than or equal to 90 mean "alpha"
     releases.

  version_semantics kernel
     Even major numbers correspond to "stable" releases.  Odd major
     numbers correspond to "alpha" releases.
	  

* Fix operation of --no-absolute-filenames --make-directories
* Restore access and modification times of symlinks in copy-in
  and copy-pass modes.
	  

Initial version.
	  

	- New constructor argument 'env' allows you to supply environment
	to the command to be run.
	- Fix packaging
	  

Fix build with mailutils 3.17
	  

	- Catch fatal errors from the Net::ACME2 module.
	- Improve pound configuration parser.
	- New options to the pound source: "--comment". "--type",
	and "--listener".  The "--comment" option instructs the
	program to issue certificates for hostnames marked by
	a specific pragmatic comment.  The "--type" option
	instructs it to process hostnames defined within listeners
	of a particular type ("http" or "https").  Finally,
	"--listener" selects listener by its tag.
	  

Initial release.
	  

* New options gcide-include-tags and gcide-exclude-tags

These options control which tags will be eligible for spell-checking.
	  

* Simultaneous execution limits

It is possible to limit number of command instances that are allowed
to run simultaneosly for a particular watcher.  This is done using
the 'max-instances' statement in 'watcher' section.

* Restore the nowait default

Unless given the 'nowait' option explicitly, watchers waited for the
handler to terminate.  It is now fixed and 'nowait' is the default, as
described in the documentation.

* Fix bug in generic to system event translation

* Fix sentinel code

In some cases setting the sentinel effectively removed the original
watcher.  That happened if the full file name of the original watcher
and its directory part produced the same hash.
	  

* New gdbm_load option: --update

The --update (-U) option instructs gdbm_load to update an existing
database.

* Fix semantics of gdbm_load -r

The --replace (-r) is valid only when used together with --update.

* Use getline in gdbmtool shell.

* New function: gdbm_load_from_file_ext

In contrast to gdbm_load and gdbm_load_from_file, which derive
the value of the flag parameter for gdbm_open from the value
of their "replace" argument, this function allows the caller to
specify it explicitly.  The prototype is:

  int gdbm_load_from_file_ext (GDBM_FILE *pdbf, FILE *fp,
			       int flags, int replace,
			       int meta_mask,
			       unsigned long *line);

* Bugfixes

** Fix binary dump format for key and/or data of zero size.
   (see https://puszcza.gnu.org.ua/bugs/?565)
** Fix location tracking and recover command in gdbtool.
   (see https://puszcza.gnu.org.ua/bugs/?566)
** Fix possible buffer underflow in gdbmload.
** Ensure any padding bytes in avail_elem structure are filled with 0.
   (fixes https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1031276)
** Improve the documentation.
	  

* Fix EOF check with flex >= 2.6.1

* Fix multiple definitions of mimetypes_error_format

* Avoid runtime errors with recent versions of libmicrohttpd
	  

Drop python 2 support
	  

* New command verb: wait

Instructs genrc to wait for the program to finish.  This implies that
the program is terminated by some third-party means.  The command
exits with code 0 if the program has finished and with code 1 if stop
timeout has been reached.
	  

* Use getgrouplist call, if available

* Fix in rush-po script

* Bugfixes
	  

* Identify service instances by container ID

* Fix regexp for hostproc docker exclusion group
	  

This is a bug-fix release with minimal changes.

* texi2any
  . fix potential crash when @include is used inside a table
  . do not complain about presence of @anchor inside @item in a table
  . C source files that are generated from *.xs files are no no longer
    distributed, so xsubpp from Perl is needed to build XS modules.
  . fix bug that led to memory alignment error on SPARC
  . performance improvement and fixes for MinGW
  . test failures due to floating point rounding error fixed (observed
    on IBM POWER9 processor)

* info
  . crash when setting style to invalid value fixed
  . potential call of memcpy with null argument fixed
	  

* Bugfix: escape single characters not occurring within a class.
	  

* New configuration statement: IgnoreSRVWeight

Instructs pound to ignore weight value of an SRV record when
generating new backend from it.  Priority of the generated backend is
copied from its matrix backend.

* New configuration statement: OverrideTTL

When used with dynamic backends, instructs pound to use the supplied
value as TTL, instead of the one returned in DNS response.

* Load balancing code revisited

Removed arbitrary limit on backend priority value.  The allowed range
is now 1..65535.

Remove priority mapping for SRV-generated backends.  SRV weights are
assigned to backend priorities verbatim.

* Fix access to freed memory in session handling code.

* Improve testsuite

Check for missing perl modules and skip tests if needed.

DNS-based tests are disabled by default, due to their experimental
nature.  Use --enable-dns-tests to enable them.

The poundharness.pl script runs a self-test when invoked with the
--fakedns option, to avoid spurious test failures.