gmime2_6-2.6.23-bp152.3.21

- Modernize spec-file by calling spec-cleaner

- Add gmime2_6-utf8.patch: Remove non-ascii line breaks from source
  files. Allows to build gmime 2.6 with gtk-doc using python 3.

- Rename package to gmime2_6, for compatability with packages not
  ready to use the new gmime 3.0 branch.

- Update to version 2.6.23:
  + gmime-crypto-context: Fixed naming.

- Update to version 2.6.22:
  + Updated Mono bindings.

- Update to version 2.6.21:
  + New function to determine if a MIME part is an attachment or
    not.
  + Treat all character values >= 128 as url-safe (bgo#738583).
  + Bugs fixed: bgo#756270, bgo#760573, bgo#760681, bgo#761263.

- Add baselibs.conf:
  + produce libgmime-2_6-0-32bit package, needed by
    libxplayer-plparser18-32bit.

- Update to GNOME 3.20.2  Fate#318572

- Add --enable-mono/--disable-mono to configure call, based on the
  build condition. This avoids random rebuild failures (outside
  OBS) if mono happens to be installed on the building system.

- Update to version 2.6.20:
  + Improved performance by using a SIMD hack when scanning for the
    end of a line.
  + Don't clear the mime_part header stream unless we actually
    remove a header.
  + Only add nwritten to the total if we've written a MIME-Version
    header.
  + Unref the disposition.
  + Don't write out 2 spaces between the ':' and the first
    message-id (bgo#711305).

- Update to version 2.6.19:
  + Custom writer for References headers so that individual msgid
    tokens do not get folded (bgo#709031).
  + Initialize the mutexes earlier (bgo#708818).
  + Fixed warnings for glib >= 2.35.1.
  + Minor code/comment cleanup.

- Update to version 2.6.18:
  + Reset the content stream after writing it.
  + Fixed an ABR by passing in an inend parameter.
  + New convenience function that handles opening the file for you.
  + Fixed g_mime_message_set_reply_to to parse the reply_to string
    just like g_mime_message_set_sender() does so that the reply_to
    field can be guaranteed to be in UTF-8.
  + Updated API documentation for functions taking or returning
    strings to clarify whether the strings should be in
    UTF-8 or not.
  + Various fixups to get rid of warnings in mono/GMime.metadata.
  + Don't call g_type_init() if glib >= 2.35.1.
  + Properly deal with obsolete routing information.

- Update to version 2.6.17:
  + If no Content-Type header exists and the parent is a
    multipart/digest, default to message/rfc822 instead of
    text/plain.
  + If the Content-Type header didn't exist, use the default
    content-type values instead of assuming text/plain.
  + According to Rfc2046 Section 5.1.1, boundary markers may
    optionally be followed by lwsp (bgo#705612).
  + Fixed to properly handle a 0-byte substream at the end of the
    very last source stream (bgo#701572).

- Update to version 2.6.16:
  + rfc2047_encode: Respect the new
    GMIME_ENABLE_USE_ONLY_USER_CHARSETS initialization flag.
  + New internal function that replaces the older header_fold()
    function. This new one uses the rfc2047 tokenizer so that we
    share the same rfc2047 workaround logic in the tokenizer
    (bgo#697407).
  + Disable g_warnings unless warnings are enabled in the build.
- Add gobject-introspection-devel BuildRequires in order to build
  the introspected bindings.
- Split typelib in typelib-1_0-GMime-2_6 subpackage.

- add buildconditional for being able to build without mono

- Update to version 2.6.15:
  + gmime/gmime-filter-html.c (html_convert): Fixed to convert the
    last line of input even if it doesn't end with a \n.

- Update to version 2.6.14:
  + Fixed the logic to actually scan for the beginning of the
    mailbox.

- Update to version 2.6.13:
  + g_mime_header_iter_{first,last}: Make sure the list isn't
    empty.
  + g_mime_header_list_prepend: Prepend the header to the list,
    don't *append* it. Oops.
  + g_mime_header_list_clear: Don't forget to reset the linked list
    so that it doesn't contained free'd nodes.

- Update to version 2.6.12:
  + stream_read): Initialize presize to READ_PAD instead of
    READ_SIZE to prevent potential buffer underruns.
  + unzip_filter: Apparently Z_BUF_ERROR isn't actually an error
    unless there is input available.
  + parser_push_boundary: Don't use strcmp to check if the boundary
    is an mbox boundary, use a direct pointer comparison so that we
    can handle Content-Type boundary parameters with a value of
    "From "
  + parser_step_headers: Don't assume that the first non-header
    line we find indicates the start of the content unless we've
    already gotten some basic headers (bgo#686223).

- Update to version 2.6.11:
  + Misc fixes.
  + Bugs fixed: bgo#679388, bgo#674032.