CMFLinkChecker 2.0 RC1 (2006-05-05)

New features

• Added methods that report all links in the database, enabling the LMS to fully synchronize with its clients.
• Added a RichTextRetriever to retrievers.py, which can search for links in content types with a RichTextWidget with the name body (patch by Maurits van Rees)
• Added a member schema to interfaces.py which CMFMembers schould include to support CMFLinkChecker. This is a temporary fix, we would like to store the whole CMFLinkChecker information inside portal_linkchecker.
• Configlet: The configuration page of the retrievers was moved to a second tab. An information frame was introduced, which is sent by the LMS. It will be used to tell you if there are updates available or maintainance work needed.
• XMLRPC protocol version: With a new LMS protocol version, the linkchecker now validates the protocol version.
• CMFLinkChecker now supports PloneHelpCenter (patch by Maurits van Rees).
• Added function retrieveAllRichTextFields that can be used by all retrievers, just like the retrieveSTX and retrieveHTML functions. It finds all fields that use the RichTextWidget and extracts links from there trying both retrieveSTX and retrieveHTML (patch by Maurits van Rees).

Bugs fixed

• Reindex the portal catalog's portal_linkchecker_uid index upon crawling the site. Per-color link overviews work reliably now.
• Uninstall the linkchecker skin layer together with the product.
• Added default actions and validators in personalize_form (patch by Maurits van Rees)
• It is now possible to re-register retrievers; used to raise ValueError (fixes bug #3293, reported by Martijn Pieters)
• Using toLocalizedTime instead of toPortalTime (patch by Maurits van Rees).
• portlet_links.pt is now a Plone 2.1 portlet (patch by Maurits van Rees).
• In the tests dir it uses self.portal instead of the deprecated self.getPortal() (patch by Maurits van Rees).
• Fixed Event retriever for Plone 2.1
• database.configure will not longer fail with AttributeError when first called (reported by Martijn Pieters, bug# 3284)
• Displaying of errormessages: If an error message contains the URL, the URL is replaced by ~ (fixes bug #2403, reported by Simon Pamies)
• Fixed the UID handling. In previous versions CMFLinkChecker would generate its own uids. We now either use Archetypes' UID() or a class provided UID (see IUIDProivder). Fixes bug #3007.
• Fixed lc_linksToMe portlet with non-archetypes objects implementing UID (Fixes bug #2806, reported by Gilles Lenfant).

CMFLinkChecker 1.9.3 (2005-06-22)

New features

• Added compatibility to LinguaPlone: LinguaPlone hides documents from other languages so we added the Language=all query to all portal_catalog queries.

Bugs fixed

• Extended anti-normalization of urls for content highlighting a bit.
• Fixed removal of links: Links were never unregistered when they got removed from a document. This also resulted in weird statistics.
• Improved interoperability with AlphaFlow: reindex doesn't fail if the user doesn't have the Modify portal content permission.
• Updated french translations (provided by Nicolas Laurance)
• Crawling is not possible when no connection to the lms can be established, therefore not showing button in config let (bug #2544)
• Unassigning retrievers in configlet is possible now (bug #2576)

CMFLinkChecker 1.9.2 (2005-04-12)

New features

• Integrated Archetype support (ATDocument retriever and automatic retrieving of all BaseObjects)
• Admin can switch on/off notifications in configlet
• Added support for kupu's resolveuid method to find backreferences.

Bugs fixed

• Upgrading old Sites (pre CMFlinkChecker 1.9) should work better now (bug #2400)
• Renaming of a Plone site with installed CMFLinkChecker works now (bug #2528)
• Matured Mail notification (fixes bug #2439, reported by Simon Pamies)
• Fixed balanced score card linking to correct detail pages.
• Fixed None filter when retrieving objects from the catalog

CMFLinkChecker 1.9.1 (2005-02-24, not publicly released)

• Do not break on certain Internet Explorer versions (java script error)
• Batching of link lists
• Improved relative to absolute path mangling
• Do not break if on certain overview pages if there are no links in the system.
• Broken links portlet gets a scrollbar if it becomes too wide

CMFLinkChecker 1.9 (2005-02-02)

• 1.9 will beome 2.0 once it's stable
• Complete refactoring to integrate an external link monitoring server (lms) instead of doing checks in a Zope thread.
• Performance improvements in all overview pages
• In-content highlighting of link-state (green, orange, red, grey)
• Bi-directional capabilities added (referenced-by portlet)
• re-install via quickinstaller possible

CMFLinkChecker 1.1

• Added french translation file. (Thanks to Nicolas) - Updated Bot-String for http protocol. (Thanks to Nicolas again)

CMFLinkChecker 1.1beta1

• Fixed the LinksForAuthenticatedUser report.
• Added configuration mechanism for the retrievers. (Bugs 1241, 2020)
• Changed registration process for retrievers.
• Fixed determining the type of some content by using portal_type instead of Type(). (Bug 2019)
• Started cleaning up I18N translations and providing a clean .POT file.
• Fixed uninstalling to completely remove any traces of the linkchecker from a Plone site. Also made the installer and uninstaller a bit more verbose in the quickinstaller log.
• Added titles for all templates to show up in the breadcrumbs instead of their Ids.
• Added correct security assertions for the scripts and templates in the skin.
• Fixed small problem with FindSupport in LinkInfo.py (thanks to Ira Fuchs).

CMFLinkChecker 1.0rc8

• Refactoring of retrievers.py that holds if CMFDefault and CMFCalendar are not installed. (This may break existing instances.)
• Minor fixes in WikiRetriever code.
• Fix for mail information that destroyed LinkInfo objects
• Fixed bug that made "Crawl site" die if the catalog query returned items that returned None for getObject()
• Removed all _blank links as those proved to be not very user friendly.
• Introduced unit tests (finally) for most components.

CMFLinkChecker 1.0rc7

• Fixed the thread "staying alive" feature.
• Fixed the permission for checking links for ManagePortal to "View"
• Not comitting checker transactions when the checker failed.
• Fixed a bug making checkers die when meeting ReadConflictErrors when loading a LinkInfo object from the database. ReadConflicts are now logged and the link is rescheduled. An additional one second pause is inserted as well.
• Introduced a new field in the linkchecker configuration view that shows the last 5 URLs that have been checked.
• Removed HEAD requests for HTTP because a lot of servers (including a famous python based open source web application server) give wrong responses.
• Added a couple of Dia UML diagrams that document the dependency and behaviour of the linkchecker components.

CMFLinkChecker 1.0rc6

• Added forgotten Docstring to checkLink method.
• Some more robust exception handling in the HTTP protocol.
• Added logging for unhandled exceptions.
• Made threads run within an infinite loop for more convenient exception behaviour.

CMFLinkChecker 1.0rc5

• Scheduling period given in hours instead of seconds. Change this, if upgrading.
• Scheduling period given in hours instead of seconds. Change this, if upgrading
• Fix for bug 1747: "activate checker thread"-button returns error Buttons for thread activation are nicely disabled/greyed now.
• Made checker links release lock after quitting/dieing.
• Refactored "resolveRelativeLink" * Fix for mailto: links
  • More stable and predictive behaviour of resolving relative Links (for each component in the URl)
• Made button for starting worker threads only available if more threads can be started (defaults to 1)
• Added .pot files and danish .po files (Thanks to Anton Stonor)
• Code cleanup
• Refactored threading code
• Setting metadata for transaction, so things show up in the undo log.
• Added "crawl my site" to the control panel by popylar request.
• Fixed bug 1698 (funny character in socket error message).
• Fixed mail notifications. They were broken because runMailNotifications is returning objects not brains.
• Made portlet for displaying link status only show up for non-Anonymous users.
• Made the runMailNotification method not use catalog result brains, but real links objects. (Broken reference to metadata field)
• Updated german .po files

CMFLinkChecker 1.0rc4

• Corrected packaging mistake that forgot shorturls.py.

CMFLinkChecker 1.0rc3

• Added more details to the balanced score card: Site wide administrator can check for all documents that contain at least one link in a certain state and check for all links in a certain state.
• More I18N
• Document links now have a title attribute that display the DC::Description field.
• The balanced score card only shows colors when at least one link matches the state. Otherwise the entries remain black.
• All URLs in the overviews are links now (with target="_blank").
• All Author/Owner information fields are mailto: links now
• URL display in lc_object_status is 50 characters long now (increased from 30) but the algorithm to shorten them has been implemented in a more intelligent fashion. (See shorturls.py)

CMFLinkChecker 1.0rc2

• Fixed for Python2.3 (socket module raising exceptions)
• Fixed handling of Timeout exception in HTTP code

CMFLinkChecker 1.0rc

• Added example to support your own content types. (See examples/WikiRetriever, just copy or link this into your Products directory.)
• Changed viewing behaviour of portlet.
• Fixed relative link handling.
• Fixed security handling for getLinksForObject and findLinksForObject.

CMFLinkChecker 1.0beta

• Completed configlet for starting threads and getting status information.
• Added support for email notifications to the authors.
• Added nice icon.
• Made the default method for HTTP handling "HEAD". Fallback is "GET" if the server responds with a "405 Method not allowed"
• Added automatic update of link lists for each object when edited. This works only automatically for the builtin types.

  See Extensions/Install how to override the default actions of the CMFFormController to activate this feature for your own content types.

• Removed tabs/border from Balanced Score Card.
• Made HTTP Checker available for HTTPS. Requires SSL sockets to be available see http://www.python.org/doc/2.2/lib/module-httplib.html on how to activate them if your python doesn't support them.
• Made some changes to the httplib for better blocking behaviour, still need to have better timings in the checker thread.

  A new socket class that handles timeouts earlier doesn't DOS the worker threads by waiting one minute for a connect, but times out at 15 seconds.

• Fixed Bug in UID generation (forgot to unwrap objects and not acquire UIDs)
• Some cleanup on ILinkInformation
• Added FTP support (good bandwith usage by calling REST if the server supports it.)
• Made protocol support pluggable
• Removed redundant status column
• Added colors to overview portlet
• Fixed (un)installation
• Fixed permissions for "My Links" and "Link" tab

CMFLinkChecker 0.3beta

• Some code cleanup (legal stuff, coding style)
• Debugging output available via zLOG.BLATHER
• Introduced the automatic scheduler. You can start it by activating a scheduler thread in the ZMI (Link Checking). It assures to check every known link at least once within the given period of time as configured in the configlet.
• Introduced "configlet" to configure the automatic scheduler and default URI prefix.
• Introduced "balanced score card" that gives strategic information about the current state of links in the site.
• Introduced "My (broken) Links" which gives an author an overview on all his broken links.
• Using a UID to maintain stable references to the objects.
• Using a ZCatalog to query LinkInfo objects.
• Using an ObjectManager for the LinkInfo objects now.
• Some usability tweaks on the link list as proposed by Alexander Limi.
• Fixed UnboundLocalError in Document retriever (#821984)
• Fixed possible crash of checker thread when urlsplit didn't find a hostname
• Bug 1503: Included a modified version for prepending a default element on URLs without hostname ("relative" links). Thanks to Sebastian Stark.
• Updated documentation
• Catching errors when parsing HTML now to avoid breaking the link discovery process.

CMFLinkChecker 0.1