Changes
=======

Silva 2.3.7 (2012/05/10)
------------------------

- Various Kupu fixes for IE 7, 9 and 10, and webkit based browsers.

Silva 2.3.6 (2012/02/23)
------------------------

- Improve tools support in Kupu to support more complex widget (used
  in Silva news).

- Improve mimetype support for Files, and extension computation. Don't
  rely on Python standard library, as they have faulty defaults. Make
  possible to provide a mime.types and a magic file in order to have
  correct computation on system that provides funky defaults.

- Improve File converter to be more efficiant: you don't need to
  create a file with the Silva File content, if the Silva File content
  is already in a file on the file system.

- Compressed Silva File should not be editable even if they are text.

- Don't loose references and metadata while converting Silva File
  storage.

- Fixed type in the code for the service_messages zmi screen, so that
  enabling the service will actually enable the service.  IF you use
  the message service in Silva 2.3 (for emailing publication messages),
  you need to go back into the zmi screen and re-enable.  The screen
  showed it was enabled, but in reality it was not.

Silva 2.3.4 (2011/10/04)
------------------------

- Port a faster implementation for the sidebar service based on the
  Zope catalog.

- Small optimizations in folder listing to improve speed.

Silva 2.3.2 (2011-07-27)
------------------------

- Fix an Acquisition issue while looking for users in service Members.

- Fix image display in XSLT renderer to high resolution.

- Fix a cache issue on the Welcome page.

- Update forms to correctly display multi-choice widgets.

- Improve SMI template.

Silva 2.3.1 (2010-02-08)
------------------------

- Review content priority in the addable list.

- Add support to automatically compute file extension like for images
  to files, by adding the correct file extension to the file
  identifier.

- Add support for file encoding (gzip, bzip2) to files (this is not
  charset !).

- Improve Kupu Silva tools (Link and specially Image tool). Don't show
  the update button if you can't update / nothing changed in the Image
  tool.

- Use the new member service API MemberLookupError to report error in
  the UI while searching for users.

- Add a compatibility fix for Zope 2.12 and old stored template in the
  ZODB.

- Fix installation of the legacy layout.

- A reference box have been added in the publish tab to show which
  content use the current content.

- The reference lookup widget have been improved to be able to propose
  the user to add content. For this a new REST API permitting to list
  addable in a container have been added.

- The HTML code of the SMI have been improved to be more easily
  skinable using only CSS (remove hardcoded style for instance).

- Fix various small bugs.


Silva 2.3 (2010/12/20)
----------------------

- Improve tab preview.

- Add a NonPublishable base class. Asset inherit from it. In a folder
  listing, non publishables objects are displayed with assets.

In Kupu:

- Improve Code Source preview, and navigation out of it.

- Fix suppression of Code Source and Images in IE.

- Fix image resizing.

- Add strikethrough.

Bug fixes:

- Fix cataloging to catalog every Silva content by default.

- Fix sending unicode emails.

- Fix order of items listed in reference widgets.

- Remove unused code (like *AccessManager*).


Silva 2.3b2 (2010/10/27)
------------------------

- Members folder is now a BTreeFolder, that improve performance on sites
  with lot of users. service_members is now an utility.

- Issues with PIL usages and libmagic have been fixed to work properly
  on native Windows.

- tab access has been reimplemented in ``silva.core.smi``. The request role
  feature has been dropped. The API to add/remove member access has
  been completely revised.

- Groups support has been removed in favor of the ``silva.pas.base``
  implementation.

- Import/Export of broken references via Silva XML is now possible.

- The subscription feature moved to a different package
  ``silva.app.subscriptions``. It can now be reused in your
  application to manage email notification.

- Sidebar cache now uses ``silva.core.cache``.

- Publication actions (Approve, Reject, Withdraw, Publish) now trigger
  Zope events.

- Kupu properties tool has been rebuilt using a REST Zeam Form with
  the help of ``zeam.form.silva``. It is now much easier to integrate a
  regular edit form inside Kupu.

- User settings (change name / email address) have been rebuilt using a
  REST form with the help of ``zeam.form.silva``.

- More tests have been improved and modernized, specially functional
  tests, using ``infrae.testbrowser``. SilvaBrowser has been removed.

- More views got modernized into Zope views.

- General usability on refactored screens has been improved
  (description, access keys, translations).


Silva 2.3b1 (2010/07/16)
------------------------

- Users can select a Gravatar image to appear next to their name in Silva.

- Silva File and Images have been reviewed, HEAD and HTTP headers are now
  properly implemented. ExtFile storage is considered legacy, please migrate
  to Blob storage.

- References are used to link content: Link, Indexer, Ghosts and links, images
  in Kupu uses it to refer linked content. XML import / export have been updated
  to work with thoses.

- z3c.forms have been replaced with Zeam Forms.

- Silva now uses Python 2.6 and Zope 2.12.

- Most of Silva contents now properly triggers Zope event when being
  added/edited.

- More parts of the SMI have been replaced by Zope views and forms.

- Tests now use infrae.testing. Functional tests rely on infrae.wsgi.

Silva 2.2 (2010-03-31)
----------------------

- A simple PhotoGallery demo view have been added to use on any
  container.

- Silva Links now support again relative links.

- SilvaExternalSources is installed by default, and the toc element in
  the default created SilvaDocument have been removed.

- Deprecate Products.Silva.helpers.makeContainerFilter and makeZMIFilter
  (makeZMIFilter was moved to silva.core.conf.utils).  Fix signature checking
  in Folder.filtered_meta_types.

- kupu: the `link` input field in the link toolbox was changed to be a
  textarea.  In 2.2 releases prior to 2.2 final, the input field changed
  to a textarea on focus and back to an input field onblur, so long urls
  where easier to view.  This caused confusing 'double clicking' behavior
  in order to manually add or update a link (first to unfocus, second to update)

- The formulator lookupwindow field changed to a textarea to address the same
  concerns as the link toolbox in kupu mentioned above.

Silva 2.2b1 (2010-02-01)
------------------------

New Features
~~~~~~~~~~~~

- External Sources are now previewable in kupu.  When an external
  source with preview enabled is added to a document, the rendered
  extneral source is shown in kupu.  When that external source is
  active, or when hovering over it with the mouse, the external source
  will have a different background color, a border, and a 'lock' icon.
  This shows the extent of the external source within the document and
  also provides a hint that it is not editable.

- The reference lookup window formulator widget now supports multiple
  entries with javascript to add / remove entries.  NOTE: the default
  is to retain the old behavior (supporting one reference, for either
  type of lookup field).  If an exsiting lookup window field is
  changed to support more the onclick widget property needs to change.
  Copy the default value from a new lookupwindowfield.

- DateTime widgets in the SMI now have a 'calendar' button which
  opens a popup calendar -- an easier way to supply datetimes.

- fixed formulator lookupwindow widget to support functioning in multi-reference
  mode in a metadata set.

Bugs Fixed
~~~~~~~~~~

- uncommon (but valid) characters were not supported in the LHS
  (mailbox) side of email addresses in mailto: links in Silva objects
  using silvaxml content storage.  An example of an email address that
  did not work: firstname.o'surname@beds.ac.uk .  The link regexp was
  expanded to allow the full range of characters in the mailbox as
  specified in RFC 2822#3.4.1.

- In kupu, when an active external source (i.e. the external source tool is
  showing that ES's parameters), clicking on the same ES no longer causes
  the tool to reload (no more flicker)

- Errors when attempting to upload interlaced PNGs are now caught and
  displayed to the user.

- launchpad #400019: Silva Link: allow editing of relative / absolute
  property, or get rid of it altogether.  Silva 2.1 allowed editing of
  this property.  Silva 2.2 has removed the link type.  Now there is
  only one type of link, and it is the authors responsibility to
  ensure it is a valid relative or absolute link.

- For a possible security issue the "served by host foo" comment was removed
  from macro_index.pt

- The user clipboard 'remove user' button now returns to the correct add user
  form. (search or direct user lookup)

- tab_preview frameset now requires that the user be logged in.  Previously, if
  using form-based auth, one might see an authentication screen in both
  frames.

- update last author when documents are saved.

- IContent objects (well, those extending Products.Silva.Content.Content) are now
  cataloged.

- enable changing the html form enctype of edit views using one of the
  macro_edit macros

- #261571: AutoTOCs have a new property "show container link". When set to 'yes'
  the public view will include an <h3> above the table of contents with
  the title of the container of the AutoTOC linking to the container.  This is
  useful if the AutoTOC is a sitemap.

Ported from Silva 2.1
~~~~~~~~~~~~~~~~~~~~~

- unpublished Versions of VersionedContent objects were not unindexed from
  the catalog when the Version was removed (e.g. via the status tab).
  A subscriber was added for the IObjectWillBeRemoved event

- fixed helpers.SwitchClass (used by the folder <--> publication conversion)
  so that the ObjectManager api's are used to set the new object, keeping
  the metadata structures up to date

Silva 2.2a2 (2009-01-12)
------------------------

New Features
~~~~~~~~~~~~

- Recursive publishing actions: on the publish tab it is now possible
  to select containers.  If a folder is selected (which is transparent)
  all subobjects are also selected.  Publication types may also be selected.
  All publishing actions will now be applied to all items contained in the
  selected containers, and recursively all items within publications.

- A New "Kupu Popup" formulator field.  This field is for html content
  and pops up a "popup kupu" to provide easy wysiwyg editing.

- Redesigned link and image tools: the link and image reference fields
  now have an icon button for the 'get reference' functionality -- saving
  toolbox space -- and when a reference is present an 'edit linked reference'
  button.  When clicked for a valid silva reference, this opens up a new window
  to the edit tab of the reference.

- The link tool's input field converts to a text area when it has the focus
  to allow for easier editing of long link references.

- External Source Tool: the select external source drop-down
  now has "Select Item" as the first option, rather than the
  ES at the top of the list.  This cleans up the interface a bit.

- 101103: allowed addables functionality is now moved to Silva
  Folders.  Any IContainer extending Products.Silva.Folder can
  now have addables settings.  NOTE: the allowed_addables API
  has also changed

Bugs Fixed
~~~~~~~~~~

- Fixed Internet Explorer 7 kupu issue triggered when clicking the
  unordered list button.

- External Source Tool: the title now reverts back to "external
  source" when the tool is reset

- External Source Tool: removing an external source (via the
  remove button) now resets the tool

- Added condition in Container/tab_edit for objects that are both
  containers and content to show only one author name.

Ported from Silva 2.1
~~~~~~~~~~~~~~~~~~~~~

- Fixed issues with unicode characters in href value of links in
  Silva Documents.

- Fixed some formatting issues with buttons in the tab editor for
  containers.

- Fixed 310589: Silva AutoTOC ignores "types to list" setting
  when using silva sort order.


- Add a function sendmail which can send mail using Mailhost and
  Maildrophost in Zope 2.11 (otherwise Maildrophost is broken in Zope
  2.11).

- Display icons of members in tab_access, the user lookup screen, and
  group edition.

Silva 2.2a1 (2008-11-18)
------------------------

Features added
~~~~~~~~~~~~~~

- 261569: the new toc features made available in 2.1 are now
  accessible within documents by using the TOC code source (id:
  cs_toc).  This is part of service_codesources, and is copied into
  the Silva Root when SilvaExternalSources is installed.

- 101103: make addables functionality available within any type of
  Container, not just Publications

- Made the public page link in the tab space visible, styled as a tab.

- Silva is shipped with eggs by default.

- Silva root became a Five local site. You can activate others Five
  local site on publications. Afterwards, you will be able to add
  'local services' to these publications.

- Five (and grok) views are supported by default on content types.

- Formlib support have been added. You can create add, edit forms, and
  every others kind of SMI forms.

- Z3C Forms support have been added. You can to create add, edit
  forms, and every others kind of SMI forms.

- A new service have been added: service_customization. It let you
  customize layout-based templates. A customization marker (ZMI
  Content) let you tag your content. After you can customize a given
  template only for that marker.

- A ZMI tab called 'Customization' have been added on every Silva
  content. It let mark this content with a customization marker, which
  can be ZODB based, or file-system based.

- Preview have been refactored out to use a namespace (++preview++)
  and a layer. You can now navigate through preview versions. Specific
  preview detection code, request tag, have been removed, now you can
  do IPreviewLayer.provided(self.request) to known if you are in
  preview mode or not.

- New skin based component have been added: content provider, viewlet
  manager, and viewlet. You can defined Grok templates.

- Resources can be pushed in a static directory of a package Python,
  like in Grok. You can use resourceinclude after to include them
  directly in your skin.

- Installation of extension have been reviewed, now you can use a
  default installer to install an extension which just add few content
  types. You can after extend this installer.

- Grok is used to configure Silva content types, services, views skins
  and forms.

- ExtensionRegistry have been re-implemented to support egg
  extension. Metadata about these extensions are fetch from egg
  metadata.

- Upgrade steps support now every complex revision number types
  (2.2.4b1-dev34 for instance).

- We (again) needed to be able to have objects other than publishable
  content and assets, and so we finally decided to not force
  everything to be an asset, but to create a new interface
  INonPublishable (from which IAsset inherits) so we can get all of
  the silva objects that are not publishable content with one method
  that is not get_assets. get_assets remains, but should now only be
  used to get all the assets (i.e. files, images, etc.) in a
  folder. To get everything that is not a publishable use
  get_non_publishables instead.

Features ported from 2.1 branch
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

- User types can select which roles they are allowed to be attributed
  to.

- Users can to *localized* and only be defined for a sub-container.

- Middleground macro have been re-factored to a view, with actions on
  it as viewlets. Extensions can plug sub-screen on any tab now,
  without overriding any tab screen.

Bugs Fixed
~~~~~~~~~~

- fixed 274668: multiple query parameters on relative links broke,
  because the ampersands were not properly escaped.

- fixed 100856: add support in Bindings for rendering the edit views
  of elements with values from the request

- Fixed 101827: lookup window was always displaying the new 'place
  reference to container...' button, even when containers
  (specifically the container's meta type) wasn't in the filter list
  for the lookup window

- Fix for kupu in html transform from
  lis.append(silva.li(el.convert(context))) to
  lis.append(silva.li(silva.br(), el.convert(context)))

- Fix for list items nesting problem

Bugfixes ported from 2.1 branch
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

- Silva Documents with code sources no longer lose the parameter types
  of the codes source parameters on xmlexport.

- Make sure that pressing the tab key when inside the kupu editor will
  always perform the indent/outdent editor commands.  This was not
  working anymore on Firefox3)

- Rebuilding the catalog sometimes ran out of memory, since it was
  done in one huge transaction. Added a generator to lazily get
  objects to reindex, and then commit after every 500 objects. The
  exact number might need tweaking. (I suspect 500 is on the
  conservative side.)

- html in parameters of external sources is no longer double escaped
  on saving.

- fixed 263503: adding support for relative links with query
  parameters (e.g. /path/to/object?key=value)

- correct unauthorized error when a chief editor/manager want to use
  the access tab and he don't have any access in the Silva root and
  the service members implements the use direct lookup protocol.

- fixed a bug in tab_edit.pt where a dynamic string was being i18n-ed.

Ported from 2.0 branch
~~~~~~~~~~~~~~~~~~~~~~

- Fixed return_url when adding content from the lookup window.

- added logging to catalog rebuilds, because they can potentially take
  a very long time.

- fixed a bug on tab_access.pt where a tal: was missing

- added logging to catalog rebuilds, because they can potentially take
  a very long time.


Silva 2.1
---------

Features added
~~~~~~~~~~~~~~

- A quota subsystem has been added. It lets Managers set a quota on a
  publication (which can also be a virtual host root). Containers
  within the publication acquire the quota. The quota does not tally
  the space used by the xml of documents but counts only assets. The
  used space is tracked on every container and displayed on the
  settings screen. The quota can only be set by a Manager in the
  settings screen of the publication. The quota subsystem is off by
  default and must be activated in 'service_extensions',

- 209497: Silva Link edit tab now displays rendered http link, while
  the preview shows the redirect within the frameset.

- 100974: the preview tab in the SMI now previews using the public
  layout, with the SMI tabs still at the top.  NOTE: TOC elements in
  documents used to link to each items '/edit/tab_preview' when the
  document was viewed via it's preview tab.  This was creating issues,
  so now links in TOC elements always link to the public version.

- 101827: A) lookup window will now list containers for any type of
  lookup window.  You can click the container title to "drill down"
  into the container, rather than (or in addition to) clicking the
  container's title in the sidebar.  In the cases where previously
  containers weren't listed, the containers are now listed, but
  references cannot be placed to them.  B) a "place reference to this
  container..." button was added, so you don't have to use the sidebar
  to navigate up one level C) Immediately when the radio button is
  selected, the reference is placed and the window is closed

- Made 'keywords' metadata field an 'index field' by default

- Use lxml instead of libxml2/libxslt for XSLT rendering. xslt
  stylesheets in extensions can import doc_elements.xslt by
  using::

    <xsl:import href="silvabase:doc_elements.xslt"/>

  A custom lxml filename resolver will find the actual file.

  Note: check encoding of output.

- Refactoring functional tests, built SilvaBrowser API for basic
  Silva functional tests.

- Took out automatic install of SilvaFind when making a new
  Silva root.  Instead a new option is added in the SilvaRoot
  addform to create a Search instance.

- The css for form button controls has been refactored. There is
  now a standard class, 'button', which can be modified by
  adding a second class name, the action type which the button
  performs. These actions are: manipulator, transporter,
  executor, remover, and canceler. E.g.  class="button
  transporter".

- Since the full media export is restricted to Chief Editors and
  up, there's no reason to show the import or export buttons to
  lower level roles (since the pure xml import/export is
  deprecated).  Conditioned the buttons.

- Modernized buttons and matrix nav based on trainee
  feedback. Use a float for button alignment classes or obviate
  the spaces in html.

- Renovated the Kupu image toolbox to make it more userfriendly.
  Improved the alignments in the table toolbox.

- New "Editor Comments" Kupu Tool: displays the "internal comments"
  silva-extra metadata field if there are comments. This promotes the
  visibility of internal / editor comments so that anyone editing the
  document will see them. This tool is only visible if there are
  comments, and does not allow editing at this time.  This toolbox is
  non-collapsible, and requires a version of kupu newer than 1.4b1 to
  support non-collapsible toolboxes.

- Support for direct username assignment to clipboard. If the
  service_members service defines a method 'direct_user_lookup' that
  returns True, the user lookup UI in the access tab is changed to an
  add user to clipboard UI. This allows the adding of new users to a
  clipboard directly.

- bug 100741: when groups are to be used, the groups management screen
  also shows groups defined elsewhere (not just local to the
  container) and provides links to edit those groups.

- 100750: "Insert new object here" functionality.  When you select an
  object in a containers edit list, then select the new object to add
  from the "add new content drop-down", the new object will be placed
  at the location of the selected object in the edit list.

- 100765 caching issues with /globals: All files in /globals are now
  tied to an Accelerated HTTP Cache Manager at
  <silva_root>/service_static_cache_manager.  This cache manager sets
  the appropriate HTTP headers to enable browsers to cache these
  objects.  The default setting is to cache for one week, and since
  the content is static, also cache non-anonymous requests.  The Silva
  root needs to be refreshed in order to take advantage of this cache
  manager.  Refreshing happens automatically during Silva upgrades.

- 101351 Extension registration now allows multuple extensions in the
  depends_on argument (in the form of tuples).  The zcml
  silva:extension depends_on directive, if present, now requires
  tuples.  Additionally the extensions service now displays the
  dependencies for each extension and displays a warning next to each
  dependency that is not available.

- Significant performance improvement in converting folders to
  publications and vice-versa

- #162545 Containers should retain the last author information
  Container author information is now displayed in the parent
  container's contents tab in the "author" column, rather than "n/a".
  Container last author and last modified dates are updated when
  objects are moved, deleted, added, pasted into the container, when
  the container's metadata is modified (properties tab and settings
  sub-tab), and when the container is converted to/from a
  publication/folder.

- developers: issue 100712: tab_edit infrastructure has been
  refactoring so that each content has a tab_edit.pt rather than the
  'edit.pt/.py' method.  The 'edit.pt/.py' method has been deprecated
  so that extensions may use this release to transition.  FMI see
  doc/developer_notes.txt

- changed the External Source kupu tool title when editing a source in
  a document so it now says "es <<source title>>" and the source id is
  displayed on hover.  External Sources in documents now display the
  title inside the "<< >>" and the id when hovered over.  The
  description of the external source is also included in the kupu
  content.

- The kupu content no longer displays the keys and values of external
  source parameters.

- AutoTOCs are now more configurable, with the following new
  features:

  1. choose Silva types that AutoTOCs will list (any Silva type,
     including Assets!)

  2. changeable depth

  3. optionally display the silva-extra content_description of content

  4. optionally show the icon of each content type

  5. Specify sort order: alphabetical, reverse-alpha, and silva SMI
     order

- 100670: The ZMI add list has been uncluttered. Silva content types
  that should only be created through the SMI are no longer addable in
  the ZMI. Also, content types that only make sense within a Silva
  Root aren't addable outside one. Developers: see
  doc/developer_changes.txt for further information on how your Silva
  extensions can take advantage of this.

Features ported from 2.0 branch
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

- added support for 'elaborate' code sources, that show their add/edit
  form in a pop-up instead of in the kupu toolbox. Useful when code
  sources have a large number of variables.

- 128518 Import of non existing values for (multi)list metadata values
  gave a formulator SelectionValidator error.

- Made Silva work with ExtFile once again: we no longer configure the
  filesystem_path in the service_files, but leave it to the
  extfile.ini.

- additional behaviour to mangle.Id.new(), when passing True as the
  first argument one can suppress the ValueError raised on invalid ids
  (and always have a new id generated)

Bugs Fixed
~~~~~~~~~~

- Took out the sec_clean_roles method in Security.py: it was
  used nowhere, and it broke with anything other than standard
  zope acl_users, because it called the non API
  get_valid_userids.

- Fixed mimetype checking in SilvaDocument to make kupu editing
  work in Firefox 3

- ie 6 compatibilite fix so smi is readable in this non
  standards compliant browser.

- 238341: frontend.css.dtml is css valid now.

- When SMTP errors occur while sending notification, we log
  them, but no longer hold up the publication/approval process.

- When pdf/word conversion results in errors, the converted text
  is indexed anyway, if there is any.

- fixed attribute error in kupu editor of changed images

- 101376: adding content with the same name as catalog indexes
  can cause problems and break the content.  The catalog index
  ids are now reserved ids.

- 101780: worked around a ZCatalog bug which causes
  containers to be unindexed twice when a parent
  container is removed.

- 101464: can't place headings inside list items in kupu

- 101228: remove slide links from object lookup window, so that the
  full SMI cannot be loaded within it

- Set overflow of selects in Kupu toolboxes to hidden, preventing long
  titles from pushing the toolbox wider.

- Changed the threshold variable in upgrade.py from 500 to 50, this
  may impact speed of upgrades negatively but it seems necessary for
  for instance ExtFile migration, where 500 objects can become too
  large for a single commit.

- Setting the i18n domain for kupu htmlmacros.html to silva so toolbox
  is translated again bug 101706.

- Changed the new version button in the editor screen from an input to
  a button, to avoid css conflicts with frontend.css controls styles

- fixed 156921: Escaping image title strings.

- fixed 101240: SilvaExternalSource cannot be disabled.

- fixed 100835: image upload screen should show allowed formats

- fixed 101174: When an id is in use on the Zope level, inform users
  accordingly

- Added True switch to Image.py tag method to escape double quotes.

- Changed the 'priority' attribute of the silva:content zcml attribute
  to a float, to support more granular priorities (they way it was
  prior to Silva 2.0 zcml)

- xslt renderer: fixed an extra whitespace issue that occurred when an
  image as linked, by setting indent="no" in normal_view.xslt

- fixed 101791: problem with Link object as index of a container

- fixed 101514: Empty list elements in Kupu.  When you press tab
  twice, Kupu inserts an element in the "missed out" level
  automatically.  This empty list element appears as a bullet that
  can't be clicked on.  Now, the 'intermediate' empty list is now
  stripped out when saving.

- fixed 101411: Compilation of an XSLT stylesheet fails if an imported
  stylesheet imports another stylesheet.  A string interpolation
  method was used to insert the path to imported stylesheets.  This
  method is no longer used, and relative paths can be used instead.
  E.g. to override noramal_view.xslt from your product's
  'transform/renderer/myrender.xslt', you can do: <xsl:import
  href="../../Silva/transform/renderer/normal_view.xslt" />

- fixed 101253: Pasted ghosts have unknown author when you click
  "paste to ghost", the resultant ghost object now has author info

- fixed 101554: Ghost Version item in service_catalog is not updated
  after haunted has been modified (example 2) When documents are
  updated, all ghosts referencing them are also reindexed.

- fixed 101064: revert to previous focus box in Content Publish screen
  remains available after reverting. The revert_to_previous
  functionality has been removed from Silva, since it was obsoleted by
  "manage versions"

Bugfixes ported from 2.0 branch
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

- Formulator lines fields no longer break when values have
  quotes or unicode characters in them.

- Message feedback from the pdf/word to text converters is once
  again presented to the user. (Was inadvertently lost in the
  Z2.10 merge)

- The id of a Silva Object is no longer in its indexed fulltext,
  as that makes no sense.

- The word and pdf to text converters no longer include error
  messages in the returned converted text.

- object_lookup.py had a missing variable definition 'addables'
  for some code paths. Added this again.

- In the reference lookup, nothing is stored in the SESSION
  anymore, preventing weird heisenbugs.

- 185422: Javascript prompt to save changes in kupu when
  leaving the editor wasn't working

- silva_lookup_referer is stored in the session. If for some
  reason the session gets lost, that no longer results in an
  error.

- lookup_search conditionally converts username to unicode. The
  name could come from a form, or from code, we don't know
  which, so we have to code defensively. :(

- KeyError in tab_status_export fixed

- Links to local (i.e. not fully qualified) domain names are no
  longer incorrectly converted to relative urls in rendered
  SilvaDocuments.

- Special characters in titles are escaped in the tocrenderer
  adapter, to play nice with xslt rendering.

- Fixed the factory function calls for File and Image, which
  still were still called with the spurious 'title' arguments on
  some code paths.

- fixed WuW144: Images were not always shown correctly in kupu
  when resized.

- fixed 158381: Group objects should be removable even if the
  groups service isn't installed.

- fixed 158384: can't add groups: title error issue

- fixed 157010: anchors in links: # -> %23

- fixed wuw144: Incorrect caching of images.
                (was reverted in silva-2.0.3b1)

- fixed 150809: sidebar cache not invalidated when folder
  removed

- Adding Groups, Virtual Groups and IPGroups was broken.

- creationdate is once again set on versioned content.

- The lookup window allowed one to add content and assets to
  ghostfolders, which isn't really advisable. This is no longer
  possible.

- Fixed the thumbnailing/caching bug. When a thumbnailed image was
  resized back to 100%, a copy was being made to make sure the
  modification date headers are newer. I changed this so only the
  modifcation date is updated, but no copy of the image is being made,
  instead the original image is used.

- Removed extra <br/> tag when viewing codesources in kupu.

- fix for bug 109008: wrong tooltip when hovering over version status
  link

- importing a zexp file with the Silva root in there was failing,
  because a Zope 3 event got fired that tried to re-add 'collection'
  to 'service_metadata' while this is already there. Added a check to
  SilvaMetadata (1.0.1) that fixes this.

- fix for bug: 101651: default settings for columns / rows on table
  creation in kupu

- widget-based and xslt renderers render underlines in the same
  way with <span class="underline">

- fixed 101494: Empty "source" elements break the XSLT renderer

- fixed 101396: paragraphs in lists, tables rendered differently in
  widgets vs xslt renderers

- fixed 101567: hires image links work from kupu but not widgets
  renderers

- fixed 148863: toc rendering not XHTML compliant

- Security.sec_get_all_roles_for_userid(userid) was broken in that it
  ignored the argument. Changed it to Security.sec_get_all_roles() and
  all calls to it, which were passing in the current userid anyway.

- fixed "excessive recursion" error when users click 'new version' to
  make a new copy of a silva document to edit then click the browser's
  back button and click 'new version' again.

- adjusted the description in the add form of links, since that
  incorrectly indicated that absolute links would be checked.

- on rare occasions, the external source tool in kupu wouldn't work
  for specific documents (double-unquoting the docref)

- Ghosts and Ghostfolders no longer show editable titles in the rename
  screen.

Features added
~~~~~~~~~~~~~~

- additional behaviour to mangle.Id.new(), when passing True as the
  first argument one can suppress the ValueError raised on invalid ids
  (and always have a new id generated)

Ported from 1.6 branch
~~~~~~~~~~~~~~~~~~~~~~

- Reversal of fix for issue
  https://bugs.launchpad.net/silva/+bug/101791 "Problem with Link
  object" since the issue was fixed in Zope 2.8+ itself, and then the
  bugfix turned out to break the link urls.

- The lookup window now opens in the directory that was last opened.

- Several reserved ids such as 'home' were declared unreserved
  previously. Unfortunately it turned out that at least 'home' was
  reserved for a good reason, so to be sure re-reserved these ids,
  except for 'search'.

- In the lookupwindowfield, a previously selected value is now shown
  selected.

Silva 2.0
---------

Features added
~~~~~~~~~~~~~~

- Use layers in our tests.  This makes sure that we can use unittests
  that set up and tear down the component architecture without
  interfering with the components that were set up for the
  integrational tests using the actual Products' ZCML.

- Use `DefaultUnicodeEncodingConflictResolver` instead of the default
  `Z2UnicodeEncodingConflictResolver` for page templates.  This means
  that our page templates will not attempt to implicitly convert to
  something when a string is not unicode and it's not a string
  containing ASCII characters only.

- Removed obsolete five:traversable statements from zcml.

- PageTemplateFile was used with non-existing page templates. This
  used to work in versions prior to Zope 2.10, but failed now. Changed
  __init__.py so that these PageTemplateFile instances are not needed
  anymore.

- Use new MessageFactory from Zope 3.3 to make it compatible with Zope
  2.10

- Changed set_mapping to mapping arguments for message IDs

- Changed css styling of the preview publisher buttons, making easier
  for users to integrate the publisher into their own layout.

- If no title is given, the index is considered a 'pure' anchor, and
  will not be found by Indexers.

- XHTML compliance improvements

Bugs Fixed
~~~~~~~~~~

- A security fix in Zope 2.10.3 broke install.py.

Ported from the 1.6 branch
~~~~~~~~~~~~~~~~~~~~~~~~~~

- External sources are removed when extracting the fulltext from a
  document, since the values of the parameters are not necessarily
  meant to be public information.

Silva 1.6
---------

Features added
~~~~~~~~~~~~~~

- 2007-03-15: Containers show their modification datetime in the
  'modified' column of the contents table in edit mode.

- If a file or image asset has an extension in its id, this is
  preserved in the fullmedia export, so if the zip file is unzipped
  by hand, the file types of the assets are obvious in those cases,
  and will preview properly in operating systems that support it.

- If present, SilvaLayout will automaticly be registered in
  service_extensions if a new SilvaRoot is created. Note that just
  having the SilvaLayout product on the file system will break the
  ZODB based layout.

- The Extension Service page now has a button 'rebuild catalog' which
  will empty the silva catalog and reindex all silva content.

- Added Atom/RSS feed capabilities to arbitrary containers. To
  use this go to the 'settings' in the properties tab of a
  container, and enable the checkbox marked 'allow feeds'. After
  that, adding /rss.xml or /atom.xml after that container's url
  just works. The feeds, for now, are feeds of only the
  published Silva Documents in that container.

- IVersioning has a method get_first_publication_date, which returns
  the earliest date any version of the content was ever published.
  Useful for RSS/Atom output.

- Silva Find features: When adding a new silvaroot a SilvaFind
  instance named "search" will be placed in the root of the site.
  SilvaFind has many improvements over the SilvaSearch
  functionality, which is still available if you remove or rename the
  SilvaFind instance. For more information on SilvaFind, have a look
  at the SilvaFind README file.

- Automatic fulltext cataloging of pdf files, when the pdftotext
  command-line application is found on the path.

- Automatic fulltext cataloging of word files, when the antiword
  command-line application is found on the path.

- Automatic fulltext cataloging of textfiles using utf8 encoding,
  also the content-type header of textfiles now set the charset to utf8.

- Ghosts now override `get_modification_datetime` to get their
  modification date from the haunted object, allowing the
  `Last-Modified` header to be set in a convenient way.
  (issue 1645)

- Services may now implement
  `Products.Silva.interfaces.IInvisibleService`, which means
  that they want to be excluded from the services listing.
  (issue 1700)

- The published status of containers changed: A folder is published
  when its default document is published, not published when its
  default document isn't published, and if no default document is
  present, it is considered published if any of its contents are
  published. This cuts down on some unnecessary recursion, as well as
  giving editors a nice way to temporarily exclude trees from
  navigation, so that they can be worked on.

- Changed title fields to string fields, for easy FF fill-in

- External sources inside headers or table cells were never allowed. An
  alert box is shown now, to make this clearer. (issue 1578)

- Disabled custom kupu context menu, so that the original browser
  context menu is shown. This allows firefox users to use the firefox
  spellchecker

- Changed the lang attribute of the layout macro html tag to get the
  lang from metadata, as in: https://infrae.com/issue/silva/issue1566
  Lower cased the language codes in the metadata set, as in:
  https://infrae.com/issue/silva/issue1668

- Moved the css selectors for rollover tooltips in Kupu toolboxes
  higher in the cascade so they apply to all toolboxes. Added tooltips
  for the new anchor/index item toolbox and the abbr/acronym toolbox.

- Fixed up the Indexer:

  - Add tbody

  - Add odd/even cycling for table rows

  - Remove double defines

  - Fix title for cases where there isn't one

  - Fix display for empty indexer

  - Reduce arbitrary indenting

- Changed css styling of the preview publisher buttons, making it easier
  for users to integrate the publisher into their own layout.

- Fix for TOCs and AutoTOCS where docs that didn't have a title
  didn't appear because the link was empty (it was there). Now the
  id shows up so visitors can at least click it.

- Renovated the TOC/AutoToc rendering, replacing the ancient, legacy,
  Netscape4 compatible image-based indent, where long titles break
  onto the next, non-indented line. The new rendering uses nested
  lists, which make the rendering easier to style, and more accessible
  to boot. The lists have the class 'toc-list', and the links have the
  'toc-link' class. Added an adapter to perform the new-style
  rendering.

- The new AutoTOC adapter is now used in SilvaDocument TOC rendering,
  for both the widgets- and xslt-based renderers.

- Tocs display links with a trailing slash because there may be
  'edit/tab_preview' appended to it, but this conflicts with links
  made in the editor that are rendered without the slash. Moved the
  slash into the appendage.

- Removed get_layout_macro(.py) from the layout installation code and
  thus it won't be present in the ZODB in future new installs. This is
  legacy code from the doomed layout service development that has now
  been superceded by SilvaLayout. Existing get_layout_macro python
  scripts can be deleted.

- Updated standard_error_message with direct call of
  layout_macro.html, 404 in title, adjusted informative text.

- index items now have an 'index title' attribute in addition to an
  'anchor name' attribute. The title is used for displaying in an
  indexer. The name is used for the anchor, and thus should be a valid
  anchor name. If no title is given, the index is considered a
  'simple' anchor, and will not be found by Indexers.

Ported from 1.5 branch
~~~~~~~~~~~~~~~~~~~~~~

- Fixes for IE7 css selectors

- With a base tag in the layout template the publisher form in public
  preview 404's. The form now has a constructed action instead of a
  point.  This will work for sites that don't define a base tag too
  (issue1705).

- When there's a base tag in the layout_macro the forms in /public,
  such as the subscription form, return a 404. If the form action is
  constructed using absolute_url, it works. This should also be
  backwards compatible, working for sites that don't include a base
  tag. Tweaked formatting of the search form while at it.

- alt attribute for meta_type icons now have the name of the meta_type
  instead of 'Unknown' (Issue 1654)

- Simplified the content.html layout file to make it easier to
  understand. Moved the content call into the layout macro.
  Reformatted the comments.

- Default installations will have the html meta tags and accessibility
  links skipped in the layout macro to have better out of the box
  performance. Site managers can turn them on as needed.

- 2007-01-24: added an explicit base tag to the layout template, so
  that it no longer includes a trailing slash which zope adds for
  all containerlike objects, which Silva Documents are. This means
  that linking to anchors now works like you would expect it to.

- Changed calls from getMetdata to getMetadataValue where possible,
  resulting in a significant performance gain.

- Non Ascii characters in names or values of codesource paramaters
  no longer result in errors.

- Incorrect links entered by hand or imported broke the path adapter.
  Since this made editing the document impossible, we added some
  defensive code.

- Anchors around divs break in explorer, so they were put inside in
  the xslt renderer.

- i18n problem in views/edit/Asset/File/submit.py (Thanks Benno!)

- Solved problem with trailing slashes in relative Links. (Thanks
  Benno!)

Bugs fixed
~~~~~~~~~~

- 2007-03-15: Ghost and GhostFolder Metadata Access Handlers return
  read only metadata bindings, so that the metadata of the haunted
  original can not be accidentally written to.

- Added a base tag so all Silva screens are based in API space,
  a.k.a. ``*/edit/`` (see issue:
  https://infrae.com/issue/silva/issue1588).

- Having a base tag fixes the potential 404 in the access screen
  (issue 1588).

- Fixed AutoTOC depth feature (issue 1693).

- Fixed behavior in Internet Explorer where it's impossible to change
  the "link to" field in the image toolbox in kupu, by removing the
  onkeypress event.

- Fixed minor bugs in kupu table toolbox where the column widths field
  did not get updated (issue 1577).

- Fixed bug in scaling of images, where images with a scaling of 100%
  would return the modification date of the original hires image,
  creating problems with caching (issue 1557).

- Fix https://infrae.com/issue/silva/issue1642, where the catalog was
  indexing XML tags.

- Fixed id mangling (https://infrae.com/issue/silva/issue1686), spaces
  are no longer allowed in new Ids.

Silva 1.6b4
-----------

Bugs fixed
~~~~~~~~~~

- New version of SilvaMetadata (0.12) released which fixes a bug where
  storing of empty metadata fields/deleting metadata values was
  impossible.

Silva 1.6b3 (first public beta)
-------------------------------

Features added
~~~~~~~~~~~~~~

- Assets now have a version_status method that always returns 'public'
  which means they can be found by SilvaFind.

- Old version cleanup code was made more robust and can now be used to
  delete all old versions older than a certain date, instead of just
  all old versions.

- Silva now uses ZCML for registration of Silva extensions and
  the content types they define. See doc/extension.txt for more
  information. See also doc/developer_changes.txt.

- The Version/CatalogedVersion base class does not expect a title
  argument anymore. This title argument had been unused for a long
  time. This requires minor adjustments in some subclasses of Version
  and CatalogedVersion. For more information see
  doc/developer_changes.txt.

- Much i18n polishing.

- Made image titles required, since they provide the alt attribute.

- Added adapters so we now have a simple API to index, unindex and
  reindex content objects, folders and assets. Adapt to
  Products.Silva.adapters.ICatalogIndexable.

- Moved document info in the document status screen to an info column
  like assets have.

- Refactored/cleaned up tab_status_macros to remove code duplication.

- Changed the add screens for Files and Images so the upload (file)
  field comes first, followed by a subhead 'optional fields', because
  Authors were not seeing the tip that they don't have to fill in an
  id.

- Added a more discreet favicon for the public views.

- For the frontend,css that gets placed in the ZMI, import the
  frontend.css in globals to stop duplication but still give the end
  user the option to edit the preview typography.

- Clarified descriptions of exporting versions.

- Added the possibility of using silva lookup to filter on multiple
  meta-types.

- The import has an extra option: 'replace content' which, when
  enabled, replaces existing content objects with objects from the
  import when they have the same ids. It treats folderish strcutures
  in the same way a POSIX cp would: i.e. in the target folders are
  used if they exist, and created if not.

- Moved the set_content_type and set_cache_headers scripts into the
  new view system, to the view 'headers'.

- Switched to Kupu 1.4 (the current 'trunk' (unreleased) version).

- Moved the 'kupu/silva' directory to 'Silva/kupu' (so to the Silva
  core) to ease release management. If you develop Silva, see
  README.txt in Silva/kupu for information on what to do when you
  change kupu templates, or when Silva needs to be ported to a newer
  version of kupu.

- Renamed 'kupu' and 'kupu_silva' in the Silva root to 'service_kupu'
  and 'service_kupu_silva', respectively.

- Removed old deprecated (de)activation API. For details, see
  'doc/developer_changes.txt'.

Bugs fixed
~~~~~~~~~~

- Browser language settings was ignoring English language, meaning it
  would fall back on the first language that was non-English in the
  settings. Now made English is a language explicitly known by Silva
  and SilvaDocument. This also means English can be explicitly
  selected in the user settings view. (issue 1572)

- The user settings view was not dealing with language variants (such
  as zh_TW) correctly. It now displays these and responds to these
  properly.

- User settings view works again. This was broken as Five views were
  in use for header settings. Made header settings view registration
  wider (for everything), and made Member object five:traversable.

- Removed commented out template code for listing all email addresses.

- Fixed various i18n translations in Versioned content

- Fixed public restriction bug cfr. issue 1588 and 1596

- Fixed bug in sidebar uplinks where you couldn't jump up and out of a
  publication.

- TRANSMAP regular expression containing weird characters was not in
  use anymore in Image.py or File.py, so removed it.

- Fix for issue1516: space above kupu in news items. Other extensions
  can control the feedback space too.

- Fixed Issue1449: Siblings Sidebar should display title or id

- Fix for Firefox html edit problem in kupu

- Single paragraphs in table cells were rendered by the xslt as
  opposed to the widget renderer. They are now removed.

- XSLT renderers did not take virtual hosting situations into account
  when rendering images and links in documents.

- In 'normal' headings index items were not rendered by the XSLT
  renderers.

- Added div with css class to make public preview buttons positionable
  in public layouts (they can be behind something if layers are used),
  thus added css class for the form too, and moved local style to the
  css so the stylesheet can be overridden by user.

- The previous fix to the public preview buttons broke keeping the
  user in the same screen when making new versions, i.e. for the
  metadata screen, where when making a new version the user was
  returned to the editor, while in fact metadata needed to be
  edited. Removed a classic very ugly url.

- If a user names the Silva root *root*, it gets priority in the dtml
  namespace, so change the variable in silva.css and frontend.css to
  ``root_url``.

- Full media export/import of images did not take account of format
  scaling and cropping.

- XMLexport of SilvaDocument passed a string instead of an integer to
  ``get_tree```/``get_public_tree``, so limiting the depth of a toc
  while using xslt rendering was ignored.

- Images in tables were not shown correctly in the xslt rendering.

- The zip import of assets (as opposed to the fullmedia zip import)
  did not take into account the 'replace identical ids' option. It
  does now.

- The xslt rendering allowed empty width and height attributes on
  images, which breaks in IE. These are now caught.

- The lookup window had some problems with incorrect content type
  filtering, and a strange javascript issue, where 'document' was
  empty. Both are fixed.

- FSDTMLMethod and FSPageTemplate were imported directly from
  FileSystemSite product instead of from fssite module

- Changed mangle.Path.fromObject() so it doesn't return '' anymore
  when from and to objects are the same.

- Handle user with user id different from login properly (issue 1471)

- Fixed bug in XML export (to_xml()) of Link objects, escaping
  (entitizing) title and url properly now.

- Removed quote (probably typo) from some JS string in
  lookupwindowfield.

- There was a bug in the cleanup script listed in MAINTENANCE.txt,
  fixed.

- Shut up a security deprecation warning for PROPFIND.

Ported from 1.5 branch
~~~~~~~~~~~~~~~~~~~~~~

- Retain fragment information (#foo) on URLs when converting from URL
  to Zope path and back. This fixes issue 1592.

- User settings screen did not work when user did not have sufficient
  rights in the Silva Root. This is now fixed.

- Fixed problem in SilvaDocument that used kupu where it whasn't
  needed and caused js exceptions.

- Fix for issue 1618 where messagebox didn't appear in ie also removed
  nbsp in ``macro_index``.

- Fixed issues where ``hide_from_toc`` filters wouldn't work in ghost
  contexts.  Also removed some unused cruft from previous filtering
  implementations.

- Added a ReferenceLookupWindowField, to store references to zope
  objects instead of urls.

- Fixed issue 1608 where ``tab_edit_revoke_approval`` caused
  versioning errors.

- In the ``tab_status`` "set timings" caused an error without saving
  the changes and returned a wrong tab

- XML export rewrote the urls of links and images on export, which was
  a bad thing. They now have extra rewritten attributes, which are
  only used by the xslt rendering.

- XML export of Silvadocuments with images that have no link attribute
  was broken.

- The full media export now uses a tempfile instead of an in memory
  stringIO to build the zipfile, hopefully improving memory
  footprint.

- Added an extra note on zope.i18n.translate to
  doc/developer_changes.txt.

- Ghostfolders exported haunted content, which broke asset ids in the
  zip. Since this content was ignored on import it is no longer
  exported.

- Explicitly delete model from request in ``view_version()`` to break
  a cyclic reference that prevents model from being garbage collected.

- Add ZCatalog index upgrader to comply with an API change in Zope.

- The new lookup window was registering resources under the name
  scripts. This name is too broad and was in fact being overridden by
  a Silva extension that introduced its own scripts, causing the Silva
  core to break. Renamed this to ``silva_scripts`` in the core.

- The cropping dialog was broken, fixed.

- Fixed non-interpolation issues with message ids in the image UI.

- Fix in the public preview screen; it broke for ghosts and other
  objects that only register the 'public' view for versions.

- Small context bug fixes in the LookupWindowWidget.

- Some text related to reserved ids wasn't properly translated.

- Failure in user preferences dialog if the browser for some reason
  has no preferred languages configured whatsoever.

- The back button should now always go back to the last edit page
  visited.

- The ``tab_edit_make_copy`` script now redirects to tab_edit instead
  of returning it, fixing a problem with public preview (where
  pressing the back button sometimes caused the make_copy script to
  get triggered) and making the URL in ``tab_edit`` look better after
  creating a new copy.

- File uploads were broken due to a missing ``return_url``
  variable. Fixed.

- Fixed multi-value form element support in external sources.

- Anchor links are rendered in a non-HTML compliant but in a Silva
  backwards compatible way.

- XSLT and widgets based renderer for anchor links are identical.

- Fixed illegal attribute value for cellpadding in XSLT renderer.

- Fixed unicode error in approval.

- GhostFolders (and probably also Ghosts) could mess up SilvaLayout if
  they referred to a location with a different skin setting than the
  location in which the GhostFolder is accessed (even when just
  looking at its title). This has now been fixed.

1.5
---

Features added
~~~~~~~~~~~~~~

- Already available in beta 1, but not in our HISTORY.txt
  file. Buttons have been added to the UI to publish and close content
  more directly.

- Added the newest translations from the Launchpad Rosetta system.
  (https://launchpad.net/products/silva/+translations)

Bugs fixed
~~~~~~~~~~

- Container policies are not hardcoded anymore for container add
  forms (add folder or add publication). Instead, the
  information on what container policies are available is
  retrieved from the service where they are registered.

- A lot more places where i18n interpolation wasn't taking place
  anymore (${foo} in message texts) are fixed by using an
  explicit call to zope.i18n.translate() instead of unicode().

- Restore "install default layout" functionality for ZODB
  layout; we were too vigorous in removing layout related code
  and removed too much.

- Removed some more traces of FileSystemSite-based layout system.

- Some fixes in the refactored lookup window code.

- Some cleanups of GhostFolder code, fixing an error that
  appeared when Five views interacted with them.

- Fix in Ghost code stops another error related to Five view
  interaction with acquisition.

- Eliminated Zope 2.8.5 warnings on startup.

- Access screen can be accessed again by authors.

- Slight simplification of icon generation code which in some
  circumstances gave rise to a bug.

- get_layout_macro was stored with DOS line endings, switched to Unix
  to be in line with the rest of the source code.

- Further bugfixes in Kupu and Kupu integration, includes Kupu 1.3.5.

1.5b1
-----

Features added
~~~~~~~~~~~~~~

- Changed over Silva so it works with Zope 2.8/Five/Zope 3
  interfaces instead of old-style interfaces. See
  doc/developer_changes for more details. These interfaces were
  in SilvaLayout as marker interfaces in the past.

- Adjusted test infrastructure so it works with Five.

- Use Zope 2.8 style transactions.

- Changed over Silva so it works with the Five/Zope 3 translation
  architecture, using Five. See doc/developer_changes for more
  details.

- Some parts of the Silva UI (for instance the lookup window in the
  edit views) are now based on Zope 3 adapters and views, using Five.
  This is a first step of a long evolutionary transition for the Silva
  UI.

- Converted IViewerSecurity, IIndexable, IHaunted, IAssetData to Zope
  3 adapters.

- The silva-extra metadata set now has a language field that stores a
  ISO 639 two letter language code.

- Language setting in the user preferences is now shown as human
  readable text, not as a language code. Moreover, using Zope 3's
  locale database, names for languages are shown in the configured
  language.

- Added support for Kupu 1.3.4.

Bugs fixed
~~~~~~~~~~

- Added missing 'is_default' method on SilvaObject, it was already
  available on Content but some new functionality (closing objects
  from tab_edit of Container types) made that it was required for
  other types as well.

- Never really an official feature, the FileSystemSite-based layout
  system has been removed from the code. This means that
  service_layouts is now gone.

- Kupu now uses frontend.css instead of an almost entire copy.

- Moved 'get_extsource_url.py' out of SilvaDocument to the core to
  allow using it from other products.

1.4.1
-----

Bugs fixed
~~~~~~~~~~

- Added support for Kupu 1.3.2, in which a SilvaExternalSources
  related bug is fixed.

- Fixed unicode bug in table rendering.

- SilvaDocument's 'fulltext' method returns a list as it did in the
  past, now that Zope has been bugfixed to handle this again.

- Fallback so that interlaced PNG files now work with Silva Images,
  even though thumbnails are disabled in that case.

1.4
---

Features added
~~~~~~~~~~~~~~

- Added a JavaScript registry for onload handlers and used that to
  do some refactoring of the tab_edit pagetemplates.

- Added support for Kupu 1.3.

- Added a field to the silva-extra metadata-set: hide_from_tocs.
  By default set to do not hide, when this is set to hide on a piece
  of content, that content (and what it contains) will not show up in
  tables of content.

- Added a link_type attribute to Link (Version) objects. Adding
  relative links is now possible.

- Browsing to objects (for e.g. link and image lookups) has been
  refactored, now there's only a single implementation of the lookup
  window, and somewhat cleaner HTML.

- It is now possible to have table cells without a width attribute,
  which means you can style your tables through css entirely. To use
  this feature, give the column widths a value of '*'.

Bugs Fixed
~~~~~~~~~~

- Un-fixed a Zope-2.7.7 compatibility issue with using ZCTextIndex.
  The original fix was basically a work arround for Zope bug. This
  means Silva can no longer work properly with Zope-2.7.7.

- Updated tests/README.txt to reflect the now preferred way for
  running the tests. ZopeTestCase-0.9.8 is now required.

- When content without a maintitle was exported, it could not be
  imported again.

- Image tags in exported silva documents had their src attribute
  converted to an absolute url on export.

- In some cases the registered upgraders for the new Silva version
  would not run. This is fixed by regsitering upgraders for all Silva
  versions, even if the particular version didn't really have any
  upgrading to do.

- Saving metadata in tab_metadata did not trigger a catalog reindex
  action.

- The test_subscriptions would fail if the MaildropHost was installed.

- pathToUrl() (path.py) has been renamed to pathToUrlPath(). Fixed
  tests accordingly.

- Various issues in Kupu have been addressed.

- Fixed various unicode errors related to PTS.

- Fixed bug where En didn't appear in the language list.

- i18n cleanups.

1.3
---

Features added
~~~~~~~~~~~~~~

- All Silva css files (silva, silvaContentStyle, frontend) have been
  cleaned up, a table of contents added, and the properties have been
  reordered following W3C recommendations. They validate.

- The border color of focused form fields highlights.

- Editable metadata fields now have labels, allowing the field to be
  focused with a mouse click.

- The content workspace now has a sidebar, which displays siblings of
  the item in the current folder.

- The middleground of properties and its subscreens is now handled by
  a macro.

- A macro_nonscreen has been added, for situations where the user clicks
  to a screen that doesn't exist, e.g. there is no addables screen for
  folders, but the user can navigate to it via the sidebar.

- macro_index has been cleaned up, and doesn't try to load Kupu unless
  Kupu is needed.

- The service_subscriptions has a service icon.

Bugs fixed
~~~~~~~~~~

- Fixed a bug in the editor transformations that made that external
  source form fields couldn't contain non-ascii characters in Kupu.

- Fixed the naive inclusion of Silva extensions in view trees, which
  caused an AttributeError on service_views when a view-less
  extension, like SilvaLayout, was installed.

- Cosmetic improvements to the public Indexer rendering.

- Changed the xmlimporter to import metadata sets that are actually
  present in the import instead of iterating over all installed metadata
  sets.

- Fixed a unicode bug in the xmlimporter.

- Fixed an API incompatibility with Zope 2.7.7 ZCTextIndex (fix works
  with 2.7.6 as well).

- Suppressed the xsltrenderer's copying of silva related namespace
  attributes and declarations in rendered html.

- Fixed a bug where unknown classnames used on p-tags (usually after
  copy/paste to Kupu) confused the editor and widget renderer.

- The user clipboard no longer hits the right side of the window.

- Numerous i18n bugs have been fixed.

- The form field widths have been fixed in the csv source edit screen.

- Various bugs and the layout have been fixed in tab_status_macros.

- If a document heading is empty, an empty h2 tag is not rendered.

- Feedback in user settings (always saying the language has been
  edited, even if it wasn't) has been fixed.

- Access keys in the contents screen have been fixed, so it is now
  possible to add items using just the keyboard.

- A bug in the select_all macro js has been fixed.

- Fixed incompatibility of code source forms between editors, requires
  Kupu 1.3 (issue 1351).

- Fixed problem with paths, that made that if you'd revoke a pending
  version of an object using the button in tab_edit of that object and
  then would edit it with Kupu, saving would fail (issue 1345).

- Fixed hyperlink support in headings in Kupu (issue 1324).

- Fixed the request approval button (issue 1337).

- Removed 'publish now' button from the public preview screen for
  authors (who are not allowed to publish anything) (issue 1369).

- Fixed image cropping button (issue 921).

- Fixed charset problems in public/subscriptor view (issue 1357).

- Fixed modification time in tab_status' version management table (issue
  1355).

- Fixed SilvaMetadata tests (XML validation problem) (issue 1354).

- Fixed AttributeError (__call__) on DummyMessageID (issue 1342).

- Fixed some bugs in Image, when an image was stored on the FS and
  removed, a Silva Image object referring to it didn't work anymore
  (issue 1350).

- Added 'X-Is-Fallback-For' directive in the appropriate .po files
  (providing information about fallbacks for e.g. de-ch to de)
  (issue 1305).

- After a cut/paste action, the paste buttons now (correctly)
  disappear from view (issue 1326).

- Fixed problems with IE mangling URLs (hrefs of anchors and src for
  images), by storing an additional attribute and reading that if
  available. Requires Kupu 1.2.2.

- Fixed bug in exporting ZODB-based files to the filesystem (for
  ExtFile), which also fixed a unit test (issue 1384).

1.3b3
-----

Note: there were a couple internal betas, this is the first public one.

Features added
~~~~~~~~~~~~~~

- WebDAV support for Silva. Silva can now be accessed by a WebDAV
  client.

- User interface refactoring for the properties (metadata) tab, and
  settings tab.

- Silva users can now set their preferred language on the user
  settings page if PTS is installed. The preference is stored in a
  cookie which expires after one year.

- Partial upgrades, an feature that allows upgrading imported
  Z-exported Silva content. This feature should be handled with care,
  as it cannot always be guaranteed to be correct.

- Public preview mode now has a back button to go back to the
  Silva UI. For EXISTING SILVA SETUPS, the preview_html Python
  script in the Silva root needs to be edited so it reads:
  return context.index_html(view_method='public_preview')
  (change 'preview' to 'public_preview').

- Metadata sets can have a 'category' property. This property
  can be used to define special purpose metadata sets.

- The minimal_role and category properties are exported and
  imported to/from the XML metadata set definitions.

- Export button in publish screen was replaced to a more logical
  location.

- The publish screen controls have been reorganized and a publish
  now button has been added, saving users a click (the publish now
  checkbox is gone).

- A publish now button has been added to almost all screens where a
  user my want to publish now, including contents, preview, public
  preview, propoerties and its subscreens.

- Users can close and publish (now) content in the contents screen.
  This makes moving of items much easier, obviating the need to switch
  to the publish screen to close.

- Both the contents and the publish screen have import and export
  (transport) buttons.

Bugs fixed
~~~~~~~~~~

- RNG schemas are up-to-date.

- Optimizations in xml export:
    * Do not use the wrappers for the dom, but the DOM directly.
    * Improved metadata values handling

- URLs generated by Indexers, now point to the correct anchors in
  the documents referenced.

- Fixed issue1246, where the cropbox didn't work correctly.

- Saving metadata uses Formulator forms for improved validation.

- The import/export of metadata values other than string now works
  correctly. (previously only datetime values were, hackishly, checked
  for.) This requires an update of Formulator.

- Fixed acquistion problem with ExternalSource objects and Kupu: when
  adding or updating an ExternalSource object that had an id that
  returned another object using acquisition, Kupu would display a 404
  page instead of the source's form. This has now been fixed: instead
  of acquisition of the id a full url is used to find the source.
  Note that this requires Kupu 1.2.1+.

- Fixed issue 1275, copy/paste and cut/paste now queries Zope for the
  ids it has generated rather than generating them in sync (which was
  broken due to a change in Zope's renaming behaviour).

- Fixed issue 1224, the 'model' variable inside the form of external
  sources now points to the document in which the code source element
  was embedded rather than to the code source itself. Note that this
  requires Kupu 1.2.1 and SilvaExternalSources 0.10.2.

- Fix unicode issues in SilvaDocument (if
  PlacelessTranslationService is used) by introducing a
  safeJoin() operation.

- Fixed issue 1277, unicode error in membership form saving mechanism.

- Fixed broken i18n:name (string interpolation) in user lookup screen.

- Fixed issue 1288, tab_status_approve should never create copies
  anymore without publishing them.

- Fixed issue 1311, some problems with nested lists in Kupu.

- Fixed issue 1301, Kupu now uses object.get_title() where appropriate.

- Updated Silva software version number to 1.3

1.2
---

Bugs fixed
~~~~~~~~~~

- batch.py now imports make_request from ZTUtils.Zope, seems to
  fix some test failures.

- layout service doesn't render pages when called from Python.

1.2b1
-----

Features added
~~~~~~~~~~~~~~

- Silva has been i18n-ed. (Works with Placeless Translation Service
  1.0). There's an English, German and Dutch version of Silva now.

- Subscriptions feature: visitors can now easily subscribe to
  publish events of VersionedContent objects (like Silva Documents).
  If a visitor is subscribed, he will receive notifications by email
  whenever a publication event for this object is triggered.

- Version management feature:

  * reviewing older versions of content objects

  * comparing different versions side-by-side

  * reverting to older versions of content

  * removal of selected old versions

- When ExtFile is installed, a site manager can switch between ZODB
  storage and file system storage of assets with one click of a
  button.

Bugs fixed
~~~~~~~~~~

- Fixed a bug in the xml exporter that tripped on the export of
  citatation elements

- Link objects now accept other protocols in the URL too (https,
  news, ftp, mailto).

- Fixed a bug in the xml importer that could result in loss of text
  data.

- Fixed color of error display when something could not be pasted.
  Note: see doc/developer_changes.txt for information about the impact
  this change had on the API.

- Removed namespace declaration from XML produced by Kupu.

- Fixed standard_error_message, it's now displayed properly when a
  user chooses 'cancel' when the browser asks for login credentials.

- Fixed problem with vanishing paragraphs after a Kupu save.

- Fixed case where XSLT renderer produced non-HTML tags.


1.1.2
-----

1.1.1
-----

Features Added
~~~~~~~~~~~~~~

- Added a new xslt renderer that renders Silva Documents without a
  title at the top.

- Made kupu the default editor if available.

Bugs Fixed
~~~~~~~~~~

- TOC rendered with the XSLT renderer now use the correct context.

- Fixed a nasty unicode error in the XSLT renderer, that somehow
  slips through in Zope 2.7.2 and only breaks Silva in 2.7.3

- Fixed issue where is_published return true in situations where
  there isn't any viewable content.

- URLs are now rendered correctly in the context of a Ghost object.

- Added 'Delete objects' permission to authors and up, this was not
  used in previous Zope versions but is in 2.7.3.

- Fixed bug that made Internet Explorer generate paths to images in
  Kupu that couldn't be reached by unauthorized users (issues 1165
  and 1168).

- Updated footer of the Silva documentation template, used to mention
  incorrect Silva and Zope versions.

- Fixed bug with metadata sets with restricted access, they were not
  accessible with people who did have the required role(s) but didn't
  have them defined on the root.

- Headings are no longer incorrectly discarded by the xslt renderer.

- Code and data sources no longer incorrectly copy their parameters
  into the result html.

- Images in SilvaDocuments now get their widths, heights, aligments
  and titles in the xml export, which fixes some bugs in the xslt
  rendering.

- Tables should have a width of 100%, but this got removed in 1.0

- Nested lists only worked in the xslt renderers when the outer list
  was bulleted. Nested lists can now have any type normal lists can.

- Newlines can be used instead of <br> in the forms editor.

- Markup and linebreaks within tables is no longer removed in the xslt
  renderers.

1.1
---

Features added
~~~~~~~~~~~~~~

- Allowed access to the cleanup adapter from Python scripts.

- Streamlined the creation of XSLT renderers.

- All objects with renderers registered for them have a way to turn
  new-style rendering off explicitly for that object in particular
  (not just in renderer registry).

Bugs fixed
~~~~~~~~~~

- Replaced print statement in Indexer for a LOG entry.

- The Indexer could break when the content tree it indexed contains
  Silva Publications.

- Protect methods on the ``service_files``.

- Content publication state is now not touched anymore when imported
  using ``*.zexp`` files in the ZMI.

- zip-file XML import/export of Silva root should now work.

- Unused XSLT support cruft has been cleaned up.

- Fixed problem with very long titles and ids in lookup screens,
  they're truncated now.

- Ghosts and ghost folders should import properly with zip file
  import.

- Fixed incorrect recursive call for ``get_public_tree_helper``.

Removals
~~~~~~~~

- Remnant of SQLDataSource removed. Use SilvaExternalSources instead.
  (SQLDataSource was deprecated since version 0.9.3).

- Remnants of old archive file import code (for bulk creation of
  assets) has been removed. Use the archivefileimport adapter instead.

1.1 beta 2
----------

Features added
~~~~~~~~~~~~~~

- Revision of the XSLT support so that it can work with any object in
  Silva, not just Silva Document. The properties tab and
  service_renderer_registry also now take this into account. If you're
  upgrading from 1.1b1, please remove service_renderer_registry in the
  ZMI and do a refresh all in service_extensions to install a new one.

Bugs fixed
~~~~~~~~~~

- Remnants of old parser code in SilvaDocument (silvaparser.py,
  search.py) and all references to it have been removed. In
  EditorSupportNested the render_links method has been removed - this
  might break old extensions.

- Remove a memory leak from XSLT support.

- Some fixes for the full-media import/export system.

- The 'get link reference' buttons in the forms-based editor now
  return the new-style HTML subset references.

- The forms-based editor now highlights index items correctly in
  nested lists and tables.

- Fixed the rename form for objects with no title.

1.1 beta 1
----------

Features added
~~~~~~~~~~~~~~

- XSLT support for rendering Silva objects. See
  doc/renderers_howto.txt for more information.

- New parser for SilvaDocument forms-based editor, using a HTML subset
  instead of the old-style method using special characters.

- A new XML export/system that can also export and import asset data.

- abbr and acronym support for both forms editor and Kupu.

- Implemented get_public_tree_all(depth) method.

Bugs fixed
~~~~~~~~~~

- Silva File and Image asset use settings from the 'nearest' (by
  acquisition) service_files again.

- Do not copy over properties anymore when changing a folder to
  publication or back. Silva does not use properties anymore.

- Properties tab for Ghost Folder works again.

- Fixed bug in images that caused a ZODB transaction on every public
  view.

- Fixed kupu transformation bug that made paragraph type not come
  through on save.

- The target attribute was not taken into account in the silva xml ->
  kupu html transformation. Fixed.

- Fixed unicode issues in Groups support. This is only an issue for
  Membership extensions that allow non-ascii characters in their group
  names, for instance the ones based on LDAP. Core groups support is
  not affected by this change.

1.0 final
---------

Bugs fixed
~~~~~~~~~~

- Made the 0.9.3-1.0 Image upgrade even more forgiving, basically it
  will now not break on anything anymore, it will print a warning to
  the zLOG on every exception.

- Problem in Kupu transformations that made 'spurious' <p> elements
  without content appear on the page.

- Problem in Kupu transformations that made <div>'s get removed
  completely from a document on save: now instead of removing it it's
  contents get converted and added to the document.

- Issue 942: quotes in links around images broke the public view.

- Added refresh all to update.

- Issues 996 and 998: changed image links that pointed to the hires
  version of the image, instead of pointing to <image>/hires_image
  they now point to <image>?hires.

- Minor problem where the image edit form in the form editor didn't
  recognize a link as one to the hires version of the image.

- Added test cases for get_creation_datetime() and
  get_modification_datetime() in SilvaObject. Also re-enabled and
  revised some tests that were commented out in test_SilvaObject.

- Indexer is now updated when installation documentation.

1.0 (release candidate 4)
-------------------------

Bugs fixed
~~~~~~~~~~

- Removed problematic code from index_html Python script that
  accidentally triggered an experimental layout service.

- Fixed bug in Kupu transformations that broke tables with incomplete
  alignment information.

- Made the 0.9.3-1.0 Image upgrade a bit more forgiving, which may
  result in broken images but at least lets the upgrade continue.

1.0 (release candidate 3)
-------------------------

Bugs fixed
~~~~~~~~~~

- Fix bug where published document that is viewed anonymously is not
  accessible.

1.0 (release candidate 2)
-------------------------

Features added
~~~~~~~~~~~~~~

- Silva picks up version from version.txt in more places so that less
  code needs to be updated before each release.

Bugs fixed
~~~~~~~~~~

- Empty index elements [[]] are now longer accepted.

- Fixed bug in default layout template in accessibility
  (previous/next) links.

- Closed content is publishable in publish tab.

1.0 (release candidate 1)
-------------------------

Features added
~~~~~~~~~~~~~~

- Kupu, a WYSIWYG editor for Silva Document. This editor offers
  a load of new features:

  - full Silva integration; Kupu knows about special Silva
    elements and saves Silva XML

  - Copy and paste HTML from any source into Kupu.

  - works in Internet Explorer (5.5 and higher) and Mozilla

  Kupu was formerly known as Epoz 1.0.

- Extensive documentation released along with Silva, and can be
  automatically installed.

- All images of a Silva Image asset are stored in the filesystem if
  ExtFile is available and the corresponding option is set in
  service_files. Also better support for ExtFile's static serving was
  introduced. service_files allows to convert the storage method of
  all Silva Image assets.

- Image scaling supports "keep aspect ratio" now. You can scale with
  ``120x*`` or ``*x211`` to fix width or height.

- An Image has a thumbnail now, retrievable via ``foo.jpg?thumbnail``.

- Cropper tool - Silva now provides facilities for cropping
  images. This tool combines with the scaling possibilities to give
  authors the means to both scale and resize a high resolution image
  to web format.

- Images can be viewed without the ``/image`` at the end (issue 594),
  the high resolution version can be retrieved via ``foo.jpg?hires``.

- Supporting a newer, improved version of ExtFile.

- Performance improvements in contents screen for folders, icon
  presentation, and metadata system.

- Numerous tweaks in the UI.

- Revision of the archive file import (i.e. zipfile import)
  functionality. Mimetypes can be registered for specific Asset
  types. The importer will create Assets depending on the mimetype.

- Silva now should deliver the right mime type for HEAD requests.

Bugs fixed
~~~~~~~~~~

- When run with Python 2.3 in 2.7 (not yet recommended by Infrae for
  Silva 1.0), suppress source code encoding warnings when starting up.

- Another problem with the upgrade code for handling index elements,
  which causes a few of them to be missed, has been repaired.

- Ghost Folders now can be added by Editors (required Manager role
  previously)

- Groups are installable as specified in the UI (no more KeyError).

- The File asset behaves more like the Image asset, where the URL to
  the object now will return the actual data. This obsoletes the need
  for URL's like "path/to/some/file/download'.

- A long standing problem with occasional InvalidObjectReference
  errors when installing or refreshing Silva should now be fixed.

0.9.3.4
-------

Bugs fixed
~~~~~~~~~~

- A problem in the upgrade code handling index elements in Silva
  Documents was repaired.

0.9.3.3
-------

Bugs fixed
~~~~~~~~~~

- Smaller bugfixes in the silvaparser.

- created a new method addAddable on ExtensionRegistry. This makes it
  possible (again) to add classes to Silva without being required to
  name your class the same as the module name (which the registerClass
  method needs). You simply register your class the normal way with
  Zope, and then invoke addAddable for its meta type.

- Fixed acquisition for viewing locally defined roles on Silva
  objects. (For instance: if user1 has editor role on pub1, and not on
  pub2, viewing /pub1/pub2/foo/tab_access shouldn't show user1 as an
  editor, and now doesn't).

0.9.3.2
-------

Bugs fixed
~~~~~~~~~~

- Make render_icon create absolute URLs again for better
  cacheability.

- Fixed the silva parser where a specific portnumber in the URL would
  not get recognized.

- Improved link parser for anchors and query fragments.

0.9.3.1
-------

Bugs fixed
~~~~~~~~~~

- Improvements in the upgrade code, esp. for large databases.

- Fixed bug in displaying the "allowed" addables.

0.9.3
-----

Author visible changes
~~~~~~~~~~~~~~~~~~~~~~

- Container objects (e.g. Folder, Publications and the Silva Root) now
  have a title independent of the index document inside the
  container. The title is set during the creation of the container and
  can be changed using the rename button in the folder edit- or
  properties tab.

- "Container policies" - When creating a new container object the
  author can choose the tpye of the index object being created, or not
  to create an index object at all. The options presented can be
  extended.

- Import assets from a zipfile now creates Silva Folder for the
  zipfiles's directory structure.

- Ghosts have a 'get reference' button now.

- Silva Files now have their icon dependent on the content. I.e. you
  get a pdf icon if you uploaded a pf file. Currently there are icons
  for PDF, Word, Excel and PowerPoint.

- Added GhostFolder: While a ghost haunts a content objects a
  GhostFolder haunts a container.

- There is a new Link object for creating references to external sites
  that can become part of navigation and TOC.  This link object
  redirects to the external site when accessed.

- The Document text markup parser has been rewritten and is able to
  deal with more forms of input properly. There are a number of
  improvements to the way hyperlinks and indexables work.

- New 'annotation' paragraph type.

- Headings in Silva Documents now allow superscript, subscript and
  emphasis markup.

- Numerous UI tweaks.

Chief editor visible changes
~~~~~~~~~~~~~~~~~~~~~~~~~~~~

- Added IP Groups. This replaces the 'restriction by IP address' text
  area in tab_access.

- Added Viewer + and Viewer ++ roles for higher view
  restrictions. Redid the user interface and underlying code for
  setting this.

Site developer visible changes
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

- Deprecated SQL Data Source assets in favor of the
  SilvaExternalSources extension. Existing SQL Data Source assets
  continue to work, but no new SQL Data Sources can be added to the
  Silva tree. In future Silva versions SQL Data Source assets may stop
  working. Add an external source in the editor using 'External
  Source' (not 'External Data', which is way to add the deprecrated
  SQL Data Source assets).

- Improved virtual hosting support. Both the public screens as well as
  the Silva edit screens should behave better in the presence of
  virtual hosting. Public view caching such as for Silva Document has
  also been made virtual host aware.

- Documents now index title along with document XML contents in
  fulltext indexing.

- Error messages in public sites should now be shown within the
  context of the public template (if standard_error_message is
  replaced in the Silva root or if a new Silva root is created).

- Some infrastructure is now in place to drive workflow from an
  outside process like a cron job. This is needed to keep the catalog
  and thus search functionality up to date if publication dates are
  set in the future or expiration dates are used.

- DemoObject is removed from the core in favor of a (to be
  developed/updated) demo extension.

Technical changes
~~~~~~~~~~~~~~~~~

- Added API to membership system so that extra columns can be defined
  for lookup results by plugin membership components.

- Merged silva.css and ploneCustom.css (removed ploneCustom) to reduce
  hits and processing.

- service_utils went away. All functionality moved to mangle.py

- Checks for valid IDs are all moved to a separate module.

- Silva Document and editor functionality is split from the Silva core
  into a separate SilvaDocument product. This is still very much tied
  to the core in this release however, so changes are noted in this
  HISTORY file for the time being.

- Silva Views machinery is split out from the Silva core into
  SilvaViews product.

- New SilvaExternalSources product replacing both code element and
  SQLSource assets. This makes it possible to pass parameters to
  external sources (a feature missing in code element). They also plug
  into the document caching framework.

- Added functionality to use CMFCore DirectoryView infrastructure
  instead of FileSystemSite, through the fssite.py module. Importing
  from the CMFCore is turned off right now, though.

  All uses of FileSystemSite functionality in Silva should now
  import through this module.

- Errors that are displayed in the Silva user interface now don't
  result in ZODB database commits.

- A new Silva unit testing framework that is based on the ZopeTestCase
  system (http://www.zope.org/Members/shh/ZopeTestCase) allows much
  higher performance and less hackery for running unit tests.  See the
  README.txt in the tests directory for more information.

- publication and modification datetime support is now not dependent
  on ZODB bobobase_modification_datetime anymore but is hooked into
  the metadata system.

Bugs fixed
~~~~~~~~~~

- Fixed rare or edge-case situations in the Publish tab. UI
  improvements, bugfixes and some code optimizations.

- Renaming publishables no longer changes ordering (issue 635)

- In the Docma queues, the fullname of the job owner is now shown
  instead of the userid, which was seen as a security issue by some
  customers.

- Fixed unicode issue when deleting a table column

- XML export permissions rationalized: it is now accessible to
  people having ReadSilvaContent permission.

- Access restriction by IP used to rely on Zope Properties;
  this doesn't happen anymore since we migrated to IP groups.

- Some cleanups in what is indexed in the catalog.

- Can add a new Silva root with non-ascii (latin1) characters
  in the title.


0.9.2.6
-------

Bugs fixed
~~~~~~~~~~

- Fulltext also indexes title, not just XML content

- Fixed some unicode errors in the EmailMessageService.

- Cache is now bypassed if 'suppress_title' is in the request.

0.9.2.5
-------

Features Added
~~~~~~~~~~~~~~

- Image elements in a Silva Document now have to possibility to link
  to the hires version or other URLs with a optional "_target".

Bugs fixed
~~~~~~~~~~

- No Catalog indexes are created for Metadata fields which do not need
  indexing.

- Metadata form and acquisition didn't work properly in all cases.

- UI improvements in tab_edit and tab_publish.

- XML export of folders when the "export last version" option is
  checked, did not use correct Folder title.

- Fixed two issues with the SQLDataSource and integer fields: the code
  doesn't try to convert the contents of the field to unicode anymore
  if it's not a plain string (fails with integers) and the code
  doesn't change the contents of the field to '' anymore when it's
  something else than None (previously it checked whether it resolved
  to true instead of checking for None).

- View caching should now be more efficient as cachability
  check is now only made once for each new publish event. The
  view caching framework now also has some unit tests.

0.9.2.4
-------

Features Added
~~~~~~~~~~~~~~

- Import assets from a zipfile now creates Silva Folder for the
  zipfiles's directory structure.

Bugs fixed
~~~~~~~~~~

- Fixed bug where the widget caches were not cleared correctly upon
  deletion of a Document.

- Fixed rare or edge-case situations in the Publish tab. UI
  improvements, bugfixes and some code optimizations.

- SilvaMetadata contained a bug which actually made metadata
  cataloging fail. This has been fixed in SilvaMetadata 0.7.2, so the
  only upgrade necessary is in fact of this dependency. For
  convenience we've also done a new Silva release, but the only thing
  it contains is one new test for getting metadata.

0.9.2.3
-------

Bugs fixed
~~~~~~~~~~

- Authors can now create new document versions again. This was a
  permission setting problem. It can be fixed by going to
  service_extensions and pressing 'refresh' for the Silva core.

- Fixed two issues with the SQLDataSource and integer fields: the code
  doesn't try to convert the contents of the field to unicode anymore
  if it's not a plain string (fails with integers) and the code
  doesn't change the contents of the field to '' anymore when it's
  something else than None (previously it checked whether it resolved
  to true instead of checking for None).

- Fixed a whole host of issues related to indexing. Indexing of
  objects was both done incorrectly and the performance was too
  slow. test_catalog was added which can be used to test indexing
  correctness, and fixes were made where problems were found. Indexing
  performance was increased by:

  * Avoiding indexing something too frequently in the Silva core.

  * Improving Formulator so that TALES expressions are cached. This
    avoids a massive Python expression compilation overhead. Upgrade
    Formulator to version 1.5.0 for a massive boost in indexing
    performance.

  * Fixing Silva metadata to avoid quadratic behavior while
    indexing. Previously, for n elements that are indexed, for each
    element indexing the data for the n indexes was accessed,
    resulting in quadratic behavior while indexing (n * n values were
    accessed per object). Upgrade SilvaMetadata to version 0.7.1 for
    another boost in indexing performance.

  * Fulltext indexing of unapproved versions is avoided.

  * Old versions should now not show up in the catalog anymore.

- Fix icon rendering. It should now be faster and more browser-cache
  friendly.

- Fixed unicode issue when deleting a table column.

- A number of things fixed in 0.9.1.x have also been fixed
  for 0.9.2.

0.9.2.2
-------

Bugs Fixed
~~~~~~~~~~

- get CSS style iteration right for containers only containing
  Assets (which otherwise would break the container view).

- Add a silva root in the ploneCustom style sheet for the
  tabbar background images (which would not display if the
  silva root folder did not have 'silva' for id and be located
  directly in the zope root).

- Be less aggressive about trying to add a Unicode Whitespace
  splitter, and accept if one is already there.

- Changed check_reserved_ids() function in the content upgrade
  script from breadth-first to depth-first. This works better for
  content hierarchies where folders with an invalid id containing
  content also with invalid ids, are being renamed.

- The metadata tab wouldn't show the metadata set's title when it
  was available.

0.9.2.1
-------

Features added
~~~~~~~~~~~~~~

- Added support for ExternalEditor, Silva Files and Images can now
  be edited using your favorite editor

Bugs Fixed
~~~~~~~~~~

- Fix bug where viewing content normally restricted by an
  AccessRestriction, could be viewed via a Ghost.

- Pressing upgrade more than once now doesn't mess up object titles
  anymore

- All content types should now have the correct metadata sets setup
  for them (led to a KeyError on 'silva-extra' in all of the views
  of those content types)

0.9.2
-----

Features Added
~~~~~~~~~~~~~~

- Bulk import of files (for File Assets) and/or images (for Image
  Assets) using a zipfile.

- New metadata architecture, which provides multiple configurable
  metadata sets. Sets can defined on a per object type basis and
  uses the catalog to make this metadata searchable.

- Unicode support. All content and meta data can contain any character
  defined by the unicode standard. On upgrade you have replace or
  update your index_html set the the correct encoding; the default
  index_html does this.

- Sidebar caching. For performance reasons, the sidebar is now cached,
  making it possible to use it throughout the site without slowing
  down page rendering.

- High resolution image support. If you have PIL installed images can
  be both converted to JPEG or PNG and arbitrary scaled for web
  presentation. The high resolution image can be used for printing.

- Asset references and code element references will now be relative
  paths if the Assets are in the same container the Document is in,
  or in a subcontainer of this container. If not, the path will be
  an absolute path.

- Docma queue viewer is now on both import and export screen. On import
  only Silva XML files are shown, on export only Word documents.

- Export screen prefills email address and description.

- Single Document import/export

- Markup character escapes (entities)

- Titles are now associated to document versions, not documents
  themselves. Different versions can have different titles.

- Some XMLWidgets related cleanups to the editor.

Bugs Fixed
~~~~~~~~~~

- (SQL)Datasources can now only be placed by ChiefEditors and up
  (using a Datasource in a document is not affected by this), and
  (SQL)Datasources now work with unicode.

- The widgets cache would be cleared correctly upon Document
  deletion. This could result in inconsistent Document editor states
  if this Document has an id earlier held by a Document now deleted.

- Setting a TOC level would display one level too much.

- Fixed problems arising from allowing to create content objects which
  may shadow important internal objects (issue321) or cause unfriendly
  Zope error messages (issue189)

- A number of minor bugs, mainly in UI.

0.9.1.5
-------

Features Added
~~~~~~~~~~~~~~

- Added support for ExternalEditor, Silva Files and Images can
  now be edited using your favorite editor.

- Bulk import of files (for File Assets) and/or images (for Image
  Assets) using a zipfile. Note though, that the 091 id
  checking is less tolerant compared to the 092 id checking.

Bugs Fixed
~~~~~~~~~~

- Removed "Asset reference" asset-type stickyness.

- Fixed button for refresh queue when Docma is enabled.

- Added a find_indexers utility (accesible through
  http://server/silva/edit/service_utils/find_indexers) to find all
  Indexer objects in use.

- Help text opens in new browser window.

- Indexer -all non-versioned content actually- can now be deleted.

- Indexer objects now render the correct URLs in virtual host
  situations. Use the 'update index' in the edit tab of Indexer
  objects to get the Indexers up to date.

- Setting a TOC level would display one level too much.

- Pressing 'Export' in Document's status tab now takes a user to the
  export screen of the container so the user can indeed (as the text
  suggests) export to Docma as well as download the content.

0.9.1.4
-------

Bugs Fixed
~~~~~~~~~~

- The UI for 'Access Restriction by Role' did not properly expose the
  possibility to set this role more restrictive on a lower level. (the
  pull down menu and button would have been disabled).

- get_public_tree() would not do the right thing with the depth- level
  control argument.

0.9.1.3
-------

Features Added
~~~~~~~~~~~~~~

- Alternating CSS classes in table rows for external data elements.

Bugs Fixed
~~~~~~~~~~

- Fixed situation where authors could not move items to other
  positions in the folders.

- Fixed bug where no asset could be added in the asset lookup dialog
  when no previous assets where available in this folder.

- Images nested in tables now keep their <a href="..."> tag (if set on
  the image element).

- Ghosts of Documents containing a TOC will render this TOC with the
  Ghost's container context and not with the container context of the
  target Document.

- Fixed unwanted aquisition behaviour of Ghost paths and other
  improvements in the Ghost code.

0.9.1.2
-------

Features Added
~~~~~~~~~~~~~~

- Backport of the Hi-res images support.

- Backport of the multipe file repositories ability.

- Backport of single Document import/export.

- Asset references and code element references will now be relative
  paths if the Assets are in the same container the Document is in, or
  in a subcontainer of this container. If not, the path will be an
  absolute path.

- Docma queue viewer is now on both import and export screen. On
  import only Silva XML files are shown, on export only Word
  documents.

- Export screen prefills email address and description.

Bugs Fixed
~~~~~~~~~~

- Fixed some bugs with copy/cut and paste: when an object is not
  allowed somewhere, it's now not possible to paste it there anymore,
  and also the id of the object to be pasted is now changed until it's
  unique (in the past, a bug in creating a unique id kept the status
  of a document 'public' in some cases).

- Intermediate fix where an local roles mapping for an non-existing
  userid would display "Unknown User" and could not be deleted.

- Past to ghost of Publications exposed two bugs. Fixed now.

0.9.1.1
-------

Bugs Fixed
~~~~~~~~~~

- UI bug where the cursor would behave inconsistently when hoovering
  buttons and buttonlinks.

- Membership now supports member objects that have different ids than
  the userids. This is useful in LDAP setups where the userid may
  contain characters illegal in URLs, such as the @ sign.

- Edit on pro transformations up to date.

0.9.1
-----

Bugs Fixed
~~~~~~~~~~

- The "use selection" button would not return to tab_access just after
  assigning a role to a user.

- Changed from absolute_url(1) to getPhysicalPath() for code elements
  - this should make code elements work in a virtual host setup.

- Datasource parameters store empty strings instead of None for
  default values.

- Fixed bug where the title of a SQLDatasource Asset was not set in
  the edit tab.

- Corrected access key in Indexer tabs.

- Fixed tab_edit where Authors can't move anything and see an empty
  controls row.

- Adjusted tab_edit and tab_status access key links so they work with
  Mozilla (select lists and checkbox labels).

- Removed Mozilla incompatibility notes from help (since the problems
  are fixed).

- Updated layout_macro to match presentation preview link in
  tab_preview.

- Fixed location of closed message in tab_preview so it shows up
  regardless of which layout (public or default) gets used.  Aligned
  presentation preview link better in default layout.

- Fix where the Groups administration button was shown, even if no
  service_groups was available.

- Now refreshes all installed products before an
  "upgrade all" action.

- Fixed bug where a new connection id was not updated correctly on the
  SQL Datasource Asset

- Indexer objects now show up in public table of contents always if
  placed (before they never did).

- File assets could only be added by users with manager role; this now
  only requires an Author role.

- A memory leak exists in Zope 2.5.1 that is exposed by Silva in some
  cases when using the SMI. Zope 2.6.1 appears to fix this memory
  leak. Note though that while Zope 2.6.1 appears to work with Silva,
  we haven't done extensive testing...

- Fixed bug which cause the upgrade of all content to crash on
  CatalogedVersionedContent objects (e.g. SilvaNews NewsItems)

- Chief editors could not add Groups or Virtual groups due to
  incorrect security settings; fixed.

0.9.1 beta
----------

Features Added
~~~~~~~~~~~~~~

- Soft breaks in paragraphs (single enter).

- Small API on Silva Root (add_silva_addable_forbidden() and
  clear_silva_addables_forbidden()) for absolutely forbidding
  visibility of certain meta_types in SMI. See also IRoot.py.

- Improved tab_access user interface.

- Group and VirtualGroup objects. Can be managed by going to 'Group
  Administration' if Group product is installed. This makes group
  management a lot easier.

- Optional message service ('service_message') which will send email
  on some workflow events.

- Membership system. Different membership policies (working for
  instance with LDAP) can be plugged into Silva and enabled by placing
  a different Member Service ('service_members').

- Access request system. A new UI so that users who do not have
  permissions to access a place can request roles on objects
  (optional).

- New 'Data Source' is now enabled by default. The first
  implementation is an 'SQL Data Source' Asset. This can be used to
  import data from a relational database into a Document using the
  'external data' element.

- List titles have been removed and transformed into headings.

- 'image_path' has been renamed to 'path' in the XML.

Bugs Fixed
~~~~~~~~~~

- Fix where the "/manage" display link in the SMI tabs didn't didn't
  regard the local roles on the current object.

- Asset reference fix where it didn't work in virtual host setups.

- Fix some unicode issue related to tables (issue184)

- Silva does no longer display non-installed content types in list of
  addables (issue159).

- Modification time of assets should now be updated properly
  (issue147).

- Some characters that were removed by the Silva editor shouldn't be
  removed -- made these work (issue 176).

- New ('add') button should work properly again (issue 162).

- Invalid image references now properly display error message (issue
  204).

0.9.0.3
-------

Bugs Fixed
~~~~~~~~~~

- Merge from HEAD for add new object UI in tab-edit.

- Fix where the "/manage" display link in the SMI tabs didn't regard
  the local roles on the current object.

- Unicode fix for table rendering. Merge from HEAD branch.

- Fix for asset reference to make it works in virtual host setups.

0.9.0.2
-------

Bugs Fixed
~~~~~~~~~~

- When using LDAPUserManagement, a Attribute error on '_getOb' occured
  when adding Silva Images and -Files.

0.9.0.1
-------

Bugs Fixed
~~~~~~~~~~

- After deleting a finished DocmaServer job, you are redirected to an
  existing page.

- The (still disabled) externaldata element was registered wrong when
  enabled; fixed this (if you uncomment the code to enable it).

0.9
---

Features Added
~~~~~~~~~~~~~~

- Can change publication to folder and the reverse.

- Links in editor now can have a 'target' option.

- Images can now be hyperlinks.

- Implementation of DataSource/DataElement for testing.

- 'Code element' that can call arbitrary code included.

- Access keys for common form buttons.

- Integration of RealObjects editor (beta).

- XML can be exported and imported.

- Docma integration for importing/exporting to other document formats
  (like MS Word). This feature needs the separate Docma server (not
  yet released).

Bugs Fixed
~~~~~~~~~~

- All '__index__.py' files are now non-empty, so that WinZip actually
  unpacks them...

- Table edit bug fixed.

- Table rendering code cleanup, better use of CSS

- Memory leak fixed (really in ParsedXML and XMLWidgets).

- Fix bug with wrong element-type selectors in editor.

- Rename now provides better user feedback.

- Definition list rendering bug fixed.

- List rendering is now XHTML compliant.

- Access restriction by role UI bug fixed.

- ZMI find should now work better, thanks to bugfix in ParsedXML.

- Lookup screen now accepts searches for people with names only 2
  characters big.

- Lists with style no bullets are now rendered with p tag.

- Pasted ghosts now obtain a last author.

- Indexer is now addable by editor and up (used to be manager
  only). Added note in UI that indexer works for published content
  only. Also fixed a bug in the Indexer; it now also indexes 'index'
  (default) documents.

- Demo index document gets automatically added to new Silva root.

- 'Authorized users below this level' moved to sub screen of access,
  to make container tab_metadata scale better (this could be very slow
  on large sites).

- Tweaked LDAPUserFolder.py so it works with more recent versions of
  LDAPUserFolder.

- Fixed problem in the Access tab, where the user lookup sometimes
  lost the current selection.

0.8.6.1
-------

Bugs Fixed
~~~~~~~~~~

- RealObjects "edit-on Pro" integration improved.

- FileSystemSite should now work on Windows install.

0.8.6
-----

Features Added
~~~~~~~~~~~~~~

- Added Indexer object, which can find index entries and present them.

- Enabled nested markup in text.

- Added easier way to add references to files, images. Can now also
  reference assets not in local folder.

- Ability to open/close branches of the site. Added a new 'Viewer'
  role. Users with this role can see closed branches.

- Request for approval now implemented; users can leave a message for
  the editors.

- Added new definition list element.

- Images now have various alignment options.

- Moved views, widgets and other support folders to the
  filesystem, now using FileSystemSite.

- Created new view registry called MultiViewRegistry which can deal
  with views from multiple Silva extension packages.  View code can be
  maintained in extension packages apart from the Silva core.

- Created installation framework for Silva extension packages.
  'service_extensions' allows site managers to install new extension
  packages into a Silva install.

- Beginnings of Silva XML import infrastructure.

Bugs fixed
~~~~~~~~~~

- Changed some permissions so it's possible to display some
  information (publication datetime for instance) to end users.

- Fixed various unicode errors.

- Single column tables are now don't break.

- Better 'no bullets' list style.

- Fixed bug with extraneous list elements when too much
  whitespace was entered.

- Fixed bug with ghosts pointing to nonexisting object.

- Reenabled caching in editor.

- Export newest version for XML exports now actually works.

- Pipe characters are now not swallowed when entered in
  paragraphs, etc.

0.8.5
-----

Features Added
~~~~~~~~~~~~~~

- New tab layout.

- Short title in Silva Document and Silva Folder metadata.

- Preformatted document element.

- Access restriction proof of concept.

- Experimental editor support for nested markup. Replace
  "EditorSupport.py" with "EditorSupportNested.py" to test it.

- Added file object.

Bugs fixed
~~~~~~~~~~

- Non existing "tab" for an object, redirects to default tab
  ("tab_edit").

- Restriction on single column tables removed.

- Various character encoding related issues.

- Removed non functional button from "tab_publish".

- List title attribute migrated to title element in lists (the upgrade
  scripts from 0.8.4 to 0.8.5 will upgrade all content containing
  lists to this new XML format).

- Display an informational message, if some view is not properly
  registered.

- UI tweaks and fixes.

0.8.4
-----

Features Added
~~~~~~~~~~~~~~

- A Silva File object to store arbitrary file in either ZODB or on
  the filesystem.

- Improved Access tab user interface.

- First implementation of group access. Groups can be assigned
  local roles and managed from Silva. The groups system is only
  enabled if the optional Groups product is installed.

- Improved object rename functionality; multiple items can now
  be renamed in one go.

- More XML export options; XML can now be exported with or
  without sub publication. It is also possible to export the
  newest version of content instead of the public version.

- Refactored Silva (API) Interfaces into separate files.

Bugs Fixed
~~~~~~~~~~

- Less restrictive object id validation (underscore and period
  is allowed).

- "addables" list is now sorted.

- Paste as ghost for a folder now also ghosts the index.

- many small UI tweaks and bugfixes.

0.8.3
-----

Features Added
~~~~~~~~~~~~~~

- Beginnings of separating out the non-core objects from Silva core.

- Some utility code in EditorSupport to support multi-column
  rendering.

- Demo Object added to demonstrate how to extend Silva with new
  objects.

- ChiefEditor and up can now configure what Silva Objects are addable
  in a Publication.

- New "Reader" role, for readonly access to Silva Management Interface
  (SMI).

- Cleanup of "Add object" infrastructure.

- Cleanup of ZMI tabs for Silva and XMLWidgets specific objects.

- Logout link now works.

Bugs Fixed
~~~~~~~~~~

- Some bug fixes to make plain Content objects (non-versioned) work
  better.

- Some changes to Ghosts to make them work properly with virtual
  hosts.

- SMI (Silva Management Interface)

  * Fix and cleanup of tabs_status code (thanks to Clemens
    Klein-Robbenhaar (a.o.) for reporting).

  * Fix of bug in Publish tab that occurred when no items selected.

  Presentation Pagetemplates:

  * Images "alt" attributes now show correct title.

  * Cleanup of index_html.

  * Removed "base" tag from layout_macro.html, fixed "script" tag,
    fixed "link" tag.

  * Changed call for "view_method" in content.html (thanks to Clemens
    Klein-Robbenhaar for reporting).

  * Changed spelling for override.html (thanks to Clemens Klein-
    Robbenhaar for reporting).

  * Various (cosmetic) changes in view/preview/tab-preview layout.

0.8.2
-----

(Internal release)

0.8.1
-----

(Internal release)

0.8
---

Initial Release

- initial public release of Silva
