NEWS

-*- coding: iso-8859-1 -*-
Mailman - The GNU Mailing List Management System
Copyright (C) 1998-2018 by the Free Software Foundation, Inc.
51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA

Here is a history of user visible changes to Mailman.

2.1.29 (24-Jul-2018)

  Bug Fixes

    - Fixed the listinfo and admin overview pages that were broken by
      LP: #1780874.  (LP: #1783417)

2.1.28 (23-Jul-2018)

  Security
 
    - A content spoofing vulnerability with invalid list name messages in
      the web UI has been fixed.  CVE-2018-13796  (LP: #1780874)

  New Features

    - It is now possible to edit HTML and text templates via the web admin
      UI in a supported language other than the list's preferred_language.
      Thanks to Yasuhito FUTATSUKI.

  i18n

    - The Japanese translation has been updated by Yasuhito FUTATSUKI.

    - The German translation has been updated by Ralf Hildebrandt.

    - The Esperanto translation has been updated by Rubén Fernández Asensio.

  Bug fixes and other patches

    - The BLOCK_SPAMHAUS_LISTED_DBL_SUBSCRIBE feature added in 2.1.27 was
      not working.  This is fixed.  (LP: #1779774)

    - Escaping of HTML entities for the web UI is now done more selectively.
      (LP: #1779445)

2.1.27 (22-Jun-2018)

  Security

    - Existing protections against malicious listowners injecting evil
      scripts into listinfo pages have had a few more checks added.
      JVN#00846677/JPCERT#97432283/CVE-2018-0618

    - A few more error messages have had their values HTML escaped.
      JVN#00846677/JPCERT#97432283/CVE-2018-0618

    - The hash generated when SUBSCRIBE_FORM_SECRET is set could have been
      the same as one generated at the same time for a different list and
      IP address.  While this is not thought to be exploitable in any way,
      the generation has been changed to avoid this.  Thanks to Ralf Jung.

  New Features

    - An option has been added to bin/add_members to issue invitations
      instead of immediately adding members.  (LP: #1773064)

    - A new BLOCK_SPAMHAUS_LISTED_IP_SUBSCRIBE setting has been added to
      enable blocking web subscribes from IPv4 addresses listed in Spamhaus
      SBL, CSS or XBL.  It will work with IPv6 addresses if Python's
      py2-ipaddress module is installed.  The module can be installed via pip
      if not included in your Python.

    - Thanks to Jim Popovitch, Mailman has a new 'security' log and logs
      authentication failures to the various web CGI functions.  The logged
      data include the remote IP and can be used to automate blocking of IPs
      with something like fail2ban.  Since Mailman 2.1.14, these have returned
      an http 401 status and the information should be logged by the web
      server, but this new log makes that more convenient.  Also, the
      'mischief' log entries for 'hostile listname' noe include the remote IP
      if available.

    - Thanks to Jim Popovitch, admin notices of (un)subscribes now may give
      the source of the action.  This consists of a %(whence)s replacement
      that has been added to the admin(un)subscribeack.txt templates.  Thanks
      to Yasuhito FUTATSUKI for updating the non-English templates and help
      with internationalizing the reasons.

    - Thanks to Jim Popovitch, there is a new
      BLOCK_SPAMHAUS_LISTED_DBL_SUBSCRIBE setting to enable blocking web
      subscribes for addresses in domains listed in the Spamhaus DBL.

  i18n

    - The Japanese translation has been updated by Yasuhito FUTATSUKI.

    - The Russian translation has been updated by Danil Smirnov.

    - A partial Esperanto translation has been added.  Thanks to
      Rubén Fernández Asensio.

    - Fixed a '# -*- coding:' line in the Russian message catalog that was
      mistakenly translated to Russian.  (LP: #1777342)

  Bug fixes and other patches

    - Some messages from bin/arch were not issued in the charset of the system
      locale when DISABLE_COMMAND_LOCALE_CSET is No.  Thanks to Yasuhito
      FUTATSUKI this is now fixed.  (LP: #1768892)

    - The message displayed in the browser when accessing a Mailman CGI when
      mm_cfg.py can't be imported due to some exception other than ImportError
      has been improved.  (LP: #1760506)

    - The reimplementation of DELIVERY_RETRY_WAIT in 2.1.26 could cause extra
      dequeueing and requeueing in the out queue by OutgoingRunner.  This is
      fixed.  (LP: #1762871)

    - A Python 2.7 dependency introduced in the ToDigests handler in Mailman
      2.1.24 has been removed.  (LP: #1755317)

    - Bad values in a list's topics will no longer break everything that
      might instantiate the list.  (LP: #1754516)

    - A Python 2.7 dependency introduced with the reCAPTCHA feature in 2.1.26
      has been removed.  (LP: #1752658)

    - The reCAPTCHA feature requires JavaScript.  If JavaScript is not enabled,
      a message will be displayed on the subscribe form that JavaScript is
      required.  (LP: #1769374)

    - Quoting in the mailman-config command has been changed from double to
      single quotes to allow double-quoted parameters.  (LP: #1774986)

    - Approving a held subscription for a user with a 'different' preferred
      language no longer corrupts the results page.  (LP: #1777222)

    - An issue with garbled descriptions on listinfo and admin overview pages
      and the heading of a list's listinfo page due to incompatible character
      sets has been fixed thanks to Yasuhito FUTATSUKI.

  Miscellaneous

    - Added to the contrib directory, a script from Jim Popovitch to generate
      Sitemap files for a list's archive.

2.1.26 (04-Feb-2018)

  Security

    - An XSS vulnerability in the user options CGI could allow a crafted URL
      to execute arbitrary javascript in a user's browser.  A related issue
      could expose information on a user's options page without requiring
      login.  These are fixed.  Thanks to Calum Hutton for the report.
      CVE-2018-5950  (LP: #1747209)

  New Features

    - Thanks to David Siebörger who adapted an existing patch by Andrea
      Veri to use Google reCAPTCHA v2 there is now the ability to add
      reCAPTCHA to the listinfo subscribe form.  There are two new mm_cfg.py
      settings for RECAPTCHA_SITE_KEY and RECAPTCHA_SECRET_KEY, the values
      for which you obtain for your domain(s) from Google at
      <https://www.google.com/recaptcha/admin>.

    - Thanks to Lindsay Haisley, there is a new bin/mailman-config command
      to display various information about this Mailman version and how it
      was configured.

  i18n

    - The Japanese message catalog has been updated for added strings by
      Yasuhito FUTATSUKI.

    - The German translation of a couple of templates has been updated by
      Thomas Hochstein.

    - The Japanese translation of Defaults.py.in has been updated by
      Yasuhito FUTATSUKI.

  Bug fixes and other patches

    - Fixed an i18n bug in the reCAPTCHA feature.  (LP: #1746189)

    - Added a few more environment variables to the list of those passed
      to CGIs to support an nginx/uwsgi configuration.  (LP #1744739)

    - Mailman 2.1.22 introduced a Python 2.7 dependency that could affect
      bin/arch processing a message without a valid Date: header.  The
      dependency has been removed.  (LP: #1740543)

    - Messages held for header_filter_rules now show the matched regexp in
      the hold reason.  (LP: #1737371)

    - When updating the group and mode of a .db file with Mailman's Postfix
      integration, a missing file is ignored.  (LP: #1734162)

    - The DELIVERY_RETRY_WAIT setting is now effective.  (LP: #1729472)

2.1.25 (26-Oct-2017)

  New Features

    - The admindb held subscriptions listing now includes the date of the
      most recent request from the address.  (LP: #1697097)

  Accessibility

    - The admin Membership List now includes text for screen readers which
      identifies the function of each checkbox.  CSS is added to the page to
      visually hide the text but still allow screen readers to read it.
      Similar text has been added to some radio buttons on the admindb pages.

  i18n

    - The Russian translation has been updated by Sergey Matveev.
      (LP: #1708016)

  Bug fixes and other patches

    - Thanks to Jim Popovitch, certain failures in DNS lookups of DMARC policy
      will now result in mitigations being applied.  (LP: #1722013)

    - The default DMARC reject reason now properly replaces %(listowner)s.
      (LP: #1718962)

    - The web roster page now shows case preserved email addresses.
      (LP: #1707447)

    - Changed the SETGID wrappers to only pass those items in the environment
      that are needed by the called scripts.  (LP: #1705736)

    - Fixed MTA/Postfix.py to ensure that created aliases(.db) and
      virtual-mailman(.db) files are readable by Postfix and the .db files are
      owned by the Mailman user.  (LP: #1696066)

    - Defended against certain web attacks that cause exceptions and "we hit
      a bug" responses when POST data or query fragments contain multiple
      values for the same parameter.  (LP: #1695667)

    - The fix for LP: #1614841 caused a regression in the options CGI.  This
      has been fixed.  (LP: #1602608)

    - Added a -a option to the (e)grep commands in contrib/mmdsr to account
      for logs that may have non-ascii and be seen as binary.

    - Fixed the -V option to bin/list_lists to not show lists whose host is a
      subdomain of the given domain.  (LP: #1695610)

2.1.24 (02-Jun-2017)

  Security

    - A most likely unexploitable XSS attach that relies on the Mailman web
      server passing a crafted Host: header to the CGI environment has been
      fixed.  Apache for one is not vulnerable.  Thanks to Alqnas Eslam.

  New Features

    - There is a new RCPT_BASE64_HEADER_NAME setting.  If this is set to a
      non-empty string, that string is the name of a header that will be added
      to personalized and VERPed deliveries with value equal to the base64
      encoding of the recipient's email address.  This is intended to enable
      identification of the recipient otherwise redacted from "spam report"
      feedback loop messages.

    - cron/senddigests has a new -e/--exceptlist option to send pending
      digests for all but a named list.  (LP: #1619770)

    - The values for DEFAULT_DIGEST_FOOTER and DEFAULT_MSG_FOOTER have been
      changed to use a standard signature separator for DEFAULT_MSG_FOOTER
      and to remove the unneded line of underscores from DEFAULT_DIGEST_FOOTER.
      (LP: #266269)

  i18n

    - The Polish html templates have been recoded to use html entities
      instead of non-ascii characters.

    - The Basque (Euskara) translation has been updated by Gari Araolaza.

    - The German "details for personalize" page has been updated by
      Christian F Buser.

    - The Japanese translation has been updated by Yasuhito FUTATSUKI.

  Bug fixes and other patches

    - The list-owner@virtual.domain addresses are now added to virtual-mailman
      as they are exposed in 'list created' emails.  (LP: #1694384)

    - The 'list run by' addresses in web page footers are now just the
      list-owner address.  (LP: #1694384)

    - Changed member_verbosity_threshold from a >= test to a strictly > test
      to avoid the issue of moderating every post when the threshold = 1.
      (LP: #1693366)

    - Subject prefixing has been improved to always have a space between
      the prefix and the subject even with non-ascii in the prefix.  This
      will sometimes result in two spaces when the prefix is non-ascii but
      the subject is ascii, but this is the lesser evil.  (LP: #1525954)

    - Treat message and digest headers and footers as empty if they contain
      only whitespace.  (LP: #1673307)

    - Ensured that added message and digest headers and footers always have
      a terminating new-line.  (LP: #1670033)

    - Fixed an uncaught TypeError in the subscribe CGI.  (LP: #1667215)

    - Added recognition for a newly seen mailEnable bounce.

    - Fixed an uncaught NotAMemberError when a member is removed before a
      probe bounce for the member is returned.  (LP: #1664729)

    - Fixed a TypeError thrown in the roster CGI when called with a listname
      containing a % character.  (LP: #1661810)

    - Fixed a NameError issue in bin/add_members with
      DISABLE_COMMAND_LOCALE_CSET = yes.  (LP: #1647450)

    - The CleanseDKIM handler has been removed from OWNER_PIPELINE.  It isn't
      needed there and has adverse DMARC implications for messages to -owner
      of an anonymous list.  (LP: #1645901)

    - Fixed an issue with properly RFC 2047 encoding the display name in the
      From: header for messages with DMARC mitigations.  (LP: #1643210)

    - Fixed an issue causing UnicodeError in sending digests following a
      change of a list's preferred_language.  (LP: #1644356)

    - Enhanced the fix for race conditions in MailList().Load().  (LP: #266464)

    - Fixed a typo in Utils.py that could have resulted in a NameError in
      logging an unlikely occurrence.  (LP: #1637745)

    - Fixed a bug which created incorrect "view more members" links at the
      bottom of the admin Membership List pages.  (LP: #1637061)

    - The 2.1.23 fix for LP: #1604544 only fixed the letter links at the top
      of the Membership List.  The links at the bottom have now been fixed.

    - paths.py now adds dist-packages as well as site-packages to sys.path.
      (LP: #1621172)

    - INIT INFO has been added to the sample init.d script.  (LP: #1620121)

2.1.23 (27-Aug-2016)

  Security

    - CSRF protection has been extended to the user options page.  This was
      actually fixed by Tokio Kikuchi as part of the fix for LP: #775294 and
      intended for Mailman 2.1.15, but that fix wasn't completely merged at the
      time.  The full fix also addresses the admindb, and edithtml pages as
      well as the user options page and the previously fixed admin pages.
      Thanks to Nishant Agarwala for reporting the issue.  CVE-2016-6893
      (LP: #1614841)
 
  New Features

    - For header_filter_rules matching, RFC 2047 encoded headers, non-encoded
      headers and header_filter_rules patterns are now all decoded to unicode.
      Both XML character references of the form &#nnnn; and unicode escapes
      of the form \Uxxxx in patterns are converted to unicodes as well.  Both
      headers and patterns are normalized to 'NFKC' normal form before
      matching, but the normalization form can be set via a new NORMALIZE_FORM
      mm_cfg setting.  Also, the web UI has been updated to encode characters
      in text fields that are invalid in the character set of the page's
      language as XML character references instead of '?'.  This should help
      with entering header_filter_rules patterns to match 'odd' characters.
      This feature is experimental and is problematic for some cases where it
      is desired to have a header_filter_rules pattern with characters not in
      the character set of the list's preferred language.  For patterns
      without such characters, the only change in behavior should be because
      of unicode normalization which should improve matching.  For other
      situations such as trying to match a Subject: with CJK characters (range
      U+4E00..U+9FFF) on an English language (ascii) list, one can enter a
      pattern like '^subject:.*[&#19968;-&#40959;]' or
      '^subject:.*[\u4e00;-\u9fff;]' to match a Subject with any character in
      the range, and it will work, but depending on the actual characters and
      the browser, submitting another, even unrelated change can garble the
      original entry although this usually occurs only with ascii pages and
      characters in the range \u0080-\u00ff.  The \Uxxxx unicode escapes must
      have exactly 4 hex digits, but they are case insensitive.  (LP: #558155)

    - Thanks to Jim Popovitch REMOVE_DKIM_HEADERS can now be set to 3 to
      preserve the original headers as X-Mailman-Original-... before removing
      them.

    - Several additional templates have been added to those that can be edited
      via the web admin GUI.  (LP: #1583387)

    - SMTPDirect.py can now do SASL authentication and STARTTLS security when
      connecting to the outgoiung MTA. Associated with this are new
      Defaults.py/mm_cfg.py settings SMTP_AUTH, SMTP_USER, SMTP_PASSWD and
      SMTP_USE_TLS.  (LP: #558281)

    - There is a new Defaults.py/mm_cfg.py setting SMTPLIB_DEBUG_LEVEL which
      can be set to 1 to enable verbose smtplib debugging to Mailman's error
      log to help with debugging 'low level smtp failures'.  (LP: #1573074)

    - A list's nonmember_rejection_notice attribute will now be the default
      rejection reason for a held non-member post in addition to it's prior
      role as the reson for an automatically rejected non-member post.
      (LP: #1572330)

  i18n

    - The French translation of 'Dutch' is changed from 'Hollandais' to
      'Néerlandais' per Francis Jorissen.

    - Some German language templates that were incorrectly utf-8 encoded have
      been recoded as iso-8859-1.  (LP: #1602779)

    - Japanese translation and documentation in messages/ja has been updated by
      Yasuhito FUTATSUKI.

  Bug fixes and other patches

    - The admin Membership List letter links could be incorrectly rendered as
      Unicode strings following a search.  (LP: #1604544)

    - We no longer throw an uncaught TypeError with certain defective crafted
      POST requests to Mailman's CGIs.  (LP: #1602608)

    - Scrubber links in archives are now in the list's preferred_language
      rather than the poster's language.  (LP: #1586505)

    - Improved logging of banned subscription and address change attempts.
      (LP: #1582856)

    - In rare circumstances a list can be removed while the admin or listinfo
      CGI or bin/list_lists is running causing an uncaught MMUnknownListError
      to be thrown.  The exception is now caught and handled.  (LP: #1582532)

    - Set the Date: header in the wrapper message when from_is_list or
      dmarc_moderation_action is Wrap Message.  (LP: #1581215)

    - A site can now set DMARC_ORGANIZATIONAL_DOMAIN_DATA_URL to None or the
      null string if it wants to avoid using this.  (LP: #1578450)
 
    - The white space to the left of the admindb Logout link is no longer
      part of the link.  (LP: #1573623)

2.1.22 (17-Apr-2016)

  i18n

    - Fixed a typo in the German options.html template.  (LP: #1562408)

    - An error in the Brazilian Portugese translation of Quarterly has been
      fixed thanks to Kleber A. Benatti.

    - The Brazilian Portugese translation has been updated by Emerson Ribeiro
      de Mello.

  Bug fixes and other patches

    - All addresses in data/virtual-mailman are now properly appended with
      VIRTUAL_MAILMAN_LOCAL_DOMAIN and duplicates are not generated if the
      site list is in a virtual domain.  (LP: #1570630)

    - DMARC mitigations will now find the From: domain to the right of the
      rightmost '@' rather than the leftmost '@'.  (LP: #1568445)

    - DMARC mitigations for a sub-domain of an organizational domain will now
      use the organizational domain's sp= policy if any.  (LP: #1568398)

    - Modified NewsRunner.py to ensure that messages gated to Usenet have a
      non-blank Subject: header and when munging the Message-ID to add the
      original to References: to help with threading.  (LP: #557955)

    - Fixed the pipermail archiver to do a better job of figuring the date of
      a post when its Date: header is missing, unparseable or has an obviously
      out of range date.  This should only affect bin/arch as ArchRunner has
      code to fix dates at least if ARCHIVER_CLOBBER_DATE_POLICY has not been
      set to 0 in mm_cfg.py.  If posts have been added in the past to a list's
      archive using bin/arch and an imported mbox, running bin/arch again could
      result is some of those posts being archived with a different date.
      (LP: #1555798)

    - Fixed an issue with CommandRunner shunting a malformed message with a
      null byte in the body.  (LP: #1553888)

    - Don't collapse multipart with a single sub-part inside multipart/signed
      parts.  (LP: #1551075)

2.1.21 (28-Feb-2016)

  New Features

    - There is a new dmarc_none_moderation_action list setting and a
      DEFAULT_DMARC_NONE_MODERATION_ACTION mm_cfg.py setting to optionally
      apply Munge From or Wrap Message actions to posts From: domains that
      publish DMARC p=none.  The intent is to eliminate failure reports to
      the domain owner for messages that would be munged or wrapped if the
      domain published a stronger DMARC policy.  See the descriptions in
      Defaults.py, the web UI and the bug report for more.  (LP: #1539384)

    - Thanks to Jim Popovitch there is now a feature to automatically turn
      on moderation for a malicious list member who attempts to flood a list
      with spam.  See the details for the Privacy options ... -> Sender
      filters -> member_verbosity_threshold and member_verbosity_interval
      settings in the web admin UI and the documentation in Defaults.py for
      the DEFAULT_MEMBER_VERBOSITY_* and VERBOSE_CLEAN_LIMIT settings for
      information.

    - bin/list_members now has options to display all moderated or all
      non-moderated members.

    - There is now a mm_cfg.py setting GLOBAL_BAN_LIST which is like the
      individual list's ban_list but applies globally to all subscribe
      requests.  See the description in Defaults.py for more details.

  i18n

    - The Japanese translation has been updated by Yasuhito FUTATSUKI.

    - Also thanks to Miloslav Trmac and Yasuhito FUTATSUKI, the l10n for
      Mailman's bin/ commands has been fixed to display using the character
      set of the user's work station even when Mailman's character set for
      the language is different.  Because this has not been tested over a
      wide set of locales, there is an mm_cfg.py switch
      DISABLE_COMMAND_LOCALE_CSET to disable it if it causes problems.
      (LP: #558167)

    - The Polish translation has been updated by Stefan Plewako.

    - The German translation has been updated by Mirian Margiani and
      Bernhard Schmidt.

    - The Russian translation has been updated by Danil Smirnov.

    - Several Galician templates that were improperly encoded as iso-8859-1
      have been fixed.  (LP: #1532504)

    - The Brazilian Portugese translation has been updated by Emerson Ribeiro
      de Mello.

  Bug fixes and other patches

    - If DMARC lookup fails to find a policy, also try the Organizational
      Domain.  Associated with this is a new mm_cfg.py setting
      DMARC_ORGANIZATIONAL_DOMAIN_DATA_URL which sets the URL used to
      retrieve the data for the algorithm that computes the Organizational
      Domain.  See https://publicsuffix.org/list/ for info.  (LP: #1549420)

    - Modified contrib/mmdsr to correctly report No such list names that
      contain ".

    - User's "Acknowledge" option will now be honored for posts to anonymous
      lists.  (LP: #1546679)

    - Fixed a typo in the Non-digest options regular_exclude_ignore
      description thanks to Yasuhito FUTATSUKI.

    - DEFAULT_PASS_MIME_TYPES has been changed to accept text/plain sub-parts
      from message/rfc822 parts and multipart parts other than mixed and
      alternative and also accept pgp signatures.  This only applies to newly
      created lists and other than pgp signatures, still only accepts
      text/plain.  (LP: #1517446)

    - Modified contrib/mmdsr to report held and banned subscriptions and DMARC
      lookups in their own categories.

    - Fixed a bug that could create a garbled From: header with certain DMARC
      mitigation actions.  (LP: #1536816)

    - Treat a poster's address which matches an equivalent_domains address as
      a list member for the regular_exclude_ignore check.  (LP: #1526550)

    - Fixed an issue that sometimes left no white space following
      subject_prefix.  (LP: #1525954)

    - Vette log entries for banned subscriptions now include the source of
      the request if available.  (LP: #1525733)

    - Submitting the user options form for a user who was asynchronously
      unsubscribed would throw an uncaught NotAMemberError.  (LP: #1523273)

    - It was possible under some circumstances for a message to be shunted
      after a handler rejected or discarded it, and the handler would be
      skipped upon unshunting and the message accepted.  (LP: #1519062)

    - Posts gated to usenet will no longer have other than the target group
      in the Newsgroups: header.  (LP: #1512866)

    - Invalid regexps in *_these_nonmembers, subscribe_auto_approval and
      ban_list are now logged.  (LP: #1507241)

    - Refactored the GetPattern list method to simplify extending @listname
      syntax to new attributes in the future.  Changed Moderate.py to use the
      GetPattern method to process the *_these_nonmembers lists.

    - Changed CookHeaders to default to using space rather than tab as
      continuation_ws when folding headers.  (LP: #1505878)

    - Fixed the 'pidfile' path in the sample init.d script.  (LP: #1503422)

    - Subject prefixing could fail to collapse multiple 'Re:' in an incomming
      message if they all came after the list's subject_prefix.  This is now
      fixed.  (LP: #1496620)

    - Defended against a user submitting URLs with query fragments or POST
      data containing multiple occurrences of the same variable.
      (LP: #1496632)

    - Fixed bin/mailmanctl to check its effective rather than real uid.
      (LP: #1491187)

    - Fixed cron/gate_news to catch EOFError on opening the newsgroup.
      (LP: #1486263)

    - Fixed a bug where a delayed probe bounce can throw an AttributeError.
      (LP: #1482940)

    - If a list is not digestable an the user is not currently set to
      receive digests, the digest options will not be shown on the user's
      options page.  (LP: #1476298)

    - Improved identification of remote clients for logging and subscribe
      form checking in cases where access is via a proxy server.  Thanks to
      Jim Popovitch.  Also updated contrib/mmdsr for log change.

    - Fixed an issue with shunted messages on a list where the charset for
      the list's preferred_language had been changed from iso-8859-1 to
      utf-8 without recoding the list's description.  (LP: #1462755)

    - Mailman-Postfix integration will now add mailman@domain entries in
      data/virtual-mailman for each domain in POSTFIX_STYLE_VIRTUAL_DOMAINS
      which is a host_name of a list.  This is so the addresses which are
      exposed on admin and listinfo overview pages of virtual domains will
      be deliverable.  (LP: #1459236)

    - The vette log entry for DMARC policy hits now contains the list name.
      (LP: #1450826)

    - If SUBSCRIBE_FORM_SECRET is enabled and a user's network has a load
      balancer or similar in use the POSTing IP might not exactly match the
      GETting IP.  This is now accounted for by not requiring the last
      octet (16 bits for ipV6) to match.  (LP: #1447445)

    - DKIM-Signature:, DomainKey-Signature: and Authentication-Results:
      headers are now removed by default from posts to anonymous lists.
      (LP: #1444673)

    - The list admin web UI Mambership List search function often doesn't
      return correct results for search strings (regexps) that contain
      non-ascii characters.  This is partially fixed.  (LP: #1442298)

2.1.20 (31-Mar-2015)

  Security

    - A path traversal vulnerability has been discovered and fixed.  This
      vulnerability is only exploitable by a local user on a Mailman server
      where the suggested Exim transport, the Postfix postfix_to_mailman.py
      transport or some other programmatic MTA delivery not using aliases
      is employed.  CVE-2015-2775  (LP: #1437145)

  New Features

    - There is a new Address Change sub-section in the web admin Membership
      Management section to allow a list admin to change a list member's
      address in one step rather than adding the new address, copying settings
      and deleting the old address.  (LP: #266809)

  i18n
 
    - The Russian translation has been updated by Danil Smirnov.

    - The Polish translation has been updated by Stefan Plewako.
 
  Bug fixes and other patches

    - A LookupError in SpamDetect on a message with RFC 2047 encoded headers
      in an unknown character set is fixed.  (LP: #1427389)

    - Fixed a bug in CommandRunner that could process the second word of a
      body line as a command word and a case sensitivity in commands in
      Subject: with an Re: prefix.  (LP: #1426829)

    - Fixed a bug in CommandRunner that threw an uncaught KeyError if
      the input to the list-request address contained a command word
      terminated by a period.  (LP: #1426825)

2.2 Branch Backports (released in conjunction with 2.1.19)

  The following New Features and Bug Fixes have been in an "unofficial,
  never to be released" Mailman 2.2 branch for several years. Until now,
  they were never implemented on the official 2.1 branch because of their
  i18n impacts.  Given that there have been a number of i18n impacting
  changes due to DMARC mitigations in the last few releases, it has been
  decided to backport these as well.

  All of these changes have been running in production on several lists
  for years without problems other than untranslated strings, so they should
  be reasonably "bug free".

  New Features

    - There is a new list attribute 'subscribe_auto_approval' which is a list
      of email addresses and regular expressions matching email addresses
      whose subscriptions are exempt from admin approval.  (LP: #266609)

    - Confirmed member change of address is logged in the 'subscribe' log,
      and if admin_notify_mchanges is true, a notice is sent to the list
      owner using a new adminaddrchgack.txt template.

    - Added an 'automate' option to bin/newlist to send the notice to the
      admin without the prompt.

    - The processing of Topics regular expressions has changed. Previously the
      Topics regexp was compiled in verbose mode but not documented as such
      which caused some confusion.  Also, the documentation indicated that
      topic keywords could be entered one per line, but these entries were not
      handled properly.  Topics regexps are now compiled in non-verbose mode
      and multi-line entries are 'ored'.  Existing Topics regexps will be
      converted when the list is updated so they will continue to work.

    - Added real name display to the web roster.  (LP: #266754)


  Bug fixes and other patches

    - Changed the response to an invalid confirmation to be more generic.
      Not all confirmations are subscription requests.

    - Changed the default nonmember_rejection_notice to be more user friendly.
      (LP: #418728)

    - Added "If you are a list member" qualification to some messages from the
      options login page.  (LP: #266442)

    - Changed the 'Approve' wording in the admindbdetails.html template to
      'Accept/Approve' for better agreement with the button labels.

    - Added '(by thread)' to the previous and next message links in the
      archive to emphasize that even if you got to the message from a
      subject, date or author index, previous and next are still by thread.

2.1.19 (28-Feb-2015)

  New Features

    - The subscribe_auto_approval feature backported from the 2.2 branch and
      described above has been enhanced to accept entries of the form
      @listname to auto approve members of another list.  (LP: #1417093)

    - There is a new list attribute dmarc_wrapped_message_text and a
      DEFAULT_DMARC_WRAPPED_MESSAGE_TEXT setting to set the default for new
      lists.  This text is added to a message which is wrapped because of
      dmarc_moderation_action in a separate text/plain part that precedes the
      message/rfc822 part containing the original message.  It can be used to
      provide an explanation of why the message was wrapped or similar info.

    - There is a new list attribute equivalent_domains and a
      DEFAULT_EQUIVALENT_DOMAINS setting to set the default for new lists which
      in turn defaults to the empty string.  This provides a way to specify one
      or more groups of domains, e.g., mac.com, me.com, icloud.com, which are
      considered equivalent for validating list membership for posting and
      moderation purposes.

    - There is a new WEB_HEAD_ADD setting to specify text to be added to the
      <HEAD> section of Mailman's internally generated web pages.  This doesn't
      apply to pages built from templates, but in those cases, custom templates
      can be created.  (LP: #1409396)

    - There is a new DEFAULT_SUBSCRIBE_OR_INVITE setting.  Set this to Yes
      to make the default selection on the admin Mass Subscriptions page
      Invite rather than Subscribe.  (LP: #1404511)

    - There is a new list attribute in the Bounce processing section.
      bounce_notify_owner_on_bounce_increment if set to Yes will cause
      Mailman to notify the list owner on every bounce that increments a
      list member's score but doesn't result in a probe or disable.  There
      is a new configuration setting setting
      DEFAULT_BOUNCE_NOTIFY_OWNER_ON_BOUNCE_INCREMENT to set the default
      for new lists.  This in turn defaults to No.  (LP: #1382150)

  Changed behavior

    - Mailman's log files, request.pck files and heldmsg-* files are no
      longer created world readable to protect against access by untrusted
      local users.  Note that permissions on existing log files won't be
      changed so if you are concerned about this and don't rotate logs or
      have a logrotate process that creates new log files instead of letting
      Mailman create them, you will need to address that.  (LP: #1327404)

  Other changes

    - The Python Powered logo image has been replaced in the misc/ directory
      in the source distribution.  Depending on how you've installed these
      images, you may need to copy PythonPowered.png from the misc/ directory
      in the source or from the $prefix/icons/ installed directory to another
      location for your web server.  (LP: #1408575)

  i18n

    - The Polish translation has been updated by Stefan Plewako.

    - The Interlingua translation has been updated by Martijn Dekker.

    - The Japanese message catalog has been updated by SATOH Fumiyasu.

    - Mailman's character set for Romanian has been changed from iso-8859-2
      to utf-8 and the templates and messages recoded.  This change will
      require running 'bin/arch --wipe' on any existing Romanian language
      lists in order to recode the list's archives, and will require recoding
      any edited templates in lists/LISTNAME/ro/*, templates/DOMAIN/ro/* and
      templates/site/ro/*.  It may also require recoding any existing
      iso-8859-2 text in list attributes.  (LP: #1418735)

    - Mailman's character set for Russian has been changed from koi8-r to
      utf-8 and the templates and messages recoded.  This change will
      require running 'bin/arch --wipe' on any existing Russian language
      lists in order to recode the list's archives, and will require recoding
      any edited templates in lists/LISTNAME/ru/*, templates/DOMAIN/ru/* and
      templates/site/ru/*.  It may also require recoding any existing koi8-r
      text in list attributes.  (LP: #1418448)

    - Mailman's versions.py has been augmented to help with the above two
      character set changes.  The first time a list with preferred_language
      of Romanian or Russian is accessed or upon upgrade to this release,
      any list attributes which have string values such as description, info,
      welcome_msg, etc. that appear to be in the old character set will be
      converted to utf-8.  This is done recursively for the values (but not
      the keys) of dictionary attributes and the elements of list and tuple
      attributes.

    - The Russian message catalog and templates have been further updated by
      Danil Smirnov.

    - The Romanian message catalog has been updated.  (LP: #1415489)

    - The Russian templates have been updated by Danil Smirnov.  (LP: #1403462)

    - The Japanese translation has been updated by SATOH Fumiyasu.
      (LP: #1402989)

    - A minor change in the French translation of a listinfo subscribe form
      message has been made.  (LP: #1331194)

  Bug fixes and other patches

    - Because of privacy concerns with the 2.2 backport adding real name to
      list rosters, this is controlled by a new ROSTER_DISPLAY_REALNAME
      setting that defaults to No.  You may wish to set this to Yes in
      mm_cfg.py.

    - Organization: headers are now unconditionally removed from posts to
      anonymous lists.  Regexps in ANONYMOUS_LIST_KEEP_HEADERS weren't kept
      if the regexp included the trailing ':'.  This is fixed too.
      (LP: #1419132)

    - The admindb interface has been fixed so the the detail message body
      display doesn't lose part of a multi-byte character, and characters which
      are invalid in the message's charset are replaced rather than the whole
      body not being converted to the display charset.  (LP: #1415406)

    - Fixed a bug in bin/rmlist that would throw an exception or just fail to
      remove held message files for a list with regexp special characters in
      its name.  (LP: #1414864)

    - When applying DMARC mitigations, CookHeaders now adds the original From:
      to Cc: rather than Reply-To: in some cases to make MUA 'reply' and
      'reply all' more consistent with the non-DMARC cases.  (LP: #1407098)

    - The Subject: of the list welcome message wasn't always in the user's
      preferred language.  Fixed.  (LP: #1400988)

    - Accept email command in Subject: prefixed with Re: or similar with no
      intervening space.  (LP: #1400200)

    - Fixed a UnicodeDecodeError that could occur in the web admin interface
      if 'text' valued attributes have unicode values.  (LP: #1397170)

    - We now catch the NotAMemberError exception thrown if an authenticated
      unsubscribe is submitted from the user options page for a nonmember.
      (LP: #1390653)

    - Fixed an archiving bug that would cause messages with 'Subject: Re:'
      only to be indexed in the archives without a link to the message.
      (LP: #1388614)

    - The vette log entry for a message discarded by a handler now includes
      the list name and the name of the handler.  (LP: #558096)

    - The options CGI now rejects all but HTTP GET and POST requests.
      (LP: #1372199)

    - A list's poster password will now be accepted on an Urgent: header.
      (LP: #1371678)

    - Fixed a bug which caused a setting of 2 for REMOVE_DKIM_HEADERS to be
      ignored.  (LP: #1363278)

    - Renamed messages/sr/readme.sr to README.sr.  (LP: #1360616)

    - Moved the dmarc_moderation_action checks from the Moderate handler to
      the SpamDetect handler so that the Reject and Discard actions will be
      done before the message might be held by header_filter_rules, and the
      Wrap Message and Munge From actions will be done on messages held by
      header_filter_rules if the message is approved.  (LP: #1334450)

    - <label> tags have been added around most check boxes and radio buttons
      and their text labels in the admin and admindb web GUI so they can be
      (de)selected by clicking the text.  (LP: #266391)

    - If checking DNS for dmarc_moderation_action and DNS lookup is not
      available, log it.  (LP: #1324541)
 
    - Handle missing From: header addresses for DMARC mitigation actions.
      (LP: #1318025)

2.1.18-1 (06-May-2014)

  Bug fixes and other patches

    - A critical incompatibility between the DMARC Wrap Message action and
      Python versions older than 2.6.x for some x <= 5 existed and caused
      Wrapped message to be shunted.  This is fixed.  (LP: #1316682)

    - Sender: headers are no longer removed in from_is_list Munge From
      actions.  (LP: #1315970)

2.1.18 (03-May-2014)

  Acknowledgements

    - Thanks to Jim Popovitch and Phil Pennock for the branch that formed the
      basis of the dmarc_moderation_action feature.

    - Thanks to Franck Martin et al for the branch that formed the basis of
      the from_is_list feature.

  Dependencies

    - There is a new dependency associated with the new Privacy options ->
      Sender filters -> dmarc_moderation_action feature discussed below.
      This requires that the dnspython <http://www.dnspython.org/> package
      be available in Python.  This package can be downloaded from the above
      site or from the CheeseShop <https://pypi.python.org/pypi/dnspython/>
      or installed with pip.

  New Features

    - The from_is_list feature introduced in 2.1.16 is now unconditionally
      available to list owners.  There is also, a new Privacy options ->
      Sender filters -> dmarc_moderation_action feature which applies to list
      messages where the From: address is in a domain which publishes a DMARC
      policy of reject or possibly quarantine.  This is a list setting with
      values of Accept, Wrap Message, Munge From, Reject or Discard. There is
      a new DEFAULT_DMARC_MODERATION_ACTION configuration setting to set the
      default for this, and the list admin UI is not able to set an action
      which is 'less' than the default.  The prior ALLOW_FROM_IS_LIST setting
      has been removed and is effectively always Yes. There is a new
      dmarc_quarantine_moderation_action list setting with default set by a
      new DEFAULT_DMARC_QUARANTINE_MODERATION_ACTION configuration setting
      which in turn defaults to Yes.  The list setting can be set to No to
      exclude domains with DMARC policy of quarantine from
      dmarc_moderation_action.

      dmarc_moderation_action and from_is_list interact in the following way.
      If the message is From: a domain to which dmarc_moderation_action applies
      and if dmarc_moderation_action is other than Accept,
      dmarc_moderation_action applies to that message.  Otherwise the
      from_is_list action applies.

      Also associated with dmarc_moderation_action are configuration settings
      DMARC_RESOLVER_TIMEOUT and DMARC_RESOLVER_LIFETIME. These are described
      in more detail in Defaults.py.  There are also new vette log entries
      written when dmarc_moderation_action is found to apply to a post.

  i18n

    - Added missing <mm-digest-question-start> tag to French listinfo template.
      (LP: #1275964)

  Bug Fixes and other patches

    - Removed HTML tags from the title of a couple of rmlist.py pages because
      browsers don't render tags in the title.  (LP: #265848)

    - Most Mailman generated notices to list owners and moderators are now
      sent as Precedence: list instead of bulk.  (LP: #1313146)

    - The Reply-To: munging options weren't honored if there was no
      from_is_list action.  (LP: #1313010)

    - Changed from_is_list actions to insert the list address in Cc: if the
      list is fully personalized.  Otherwise, the list address is only in
      From: and Reply-To: overrides it.  (LP: #1312970)

    - Fixed the Munge From action to only Munge the From: and/or Reply-To: in
      the outgoing message and not in archives, digests and messages sent via
      the usenet gateway.  (LP: #1311431)

    - Fixed a long standing issue in which a notice sent to a user whose
      language is other than that of the list can cause subsequent things
      which should be in the list's language to be in the user's language