Products.Silva 2.3.7

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.

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.

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

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

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.

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.

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.

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.

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

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.

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.

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)

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”

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.

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)

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.

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

A security fix in Zope 2.10.3 broke install.py.

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.

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.

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!)

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.

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’.

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.

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.

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)

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.

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.

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.

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 ‘*’.

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.

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.

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).

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.

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

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.

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.

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

Made kupu the default editor if available.

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.

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).

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.

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.

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.

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.

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.

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.

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

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.

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.

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.

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.

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.

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.

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

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.

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

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.

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

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)

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.

(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.

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.

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.

Alternating CSS classes in table rows for external data elements.

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.

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.

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.

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.

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).

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).

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.

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.

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.

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.

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.

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.

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.

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.