Zope 2 2.6.2b5

Zope Changes

  This file contains change information for the current Zope release.
  Change information for previous versions of Zope can be found in the
  file HISTORY.txt.

  Zope 2.6.2 beta 5

    Bugs Fixed

      - ZClasses that subclassed ObjectManager that were created in
        earlier versions of Zope would not load under 2.6, due to
        the new Interfaces package. Added back a simple stub module
        and changed a constructor to allow these ZClasses to work.

      - Collector #823: XMLRPC exception string marshalling bug fix
        backported from head

  Zope 2.6.2 beta 4

    Bugs Fixed

      - Collector: #964: standard_error_message refers to looking into the
        HTML code for more information which is deprecated. Referring to
        the error log now.

      - Collector #959/#514: calling has_permission on a user object
        did not use the user object instance to check the permission;
        rather, it used the currently logged in user :(

      - Fixed a potential bug in ZTUtils.Tree.decodeExpansion where a
        potentially empty string was tested for it's first character; used
        .startswith for safety.

  Zope 2.6.2 beta 3

    Backward incompatibilities

      - We no longer honor local security settings that would allow
        someone to join or leave versions unless the location of the
        settings is a folder directly or indirectly containing the
        user's user folder.

    Bugs Fixed

      - Collector #893: Mailhost: munge_header has been broken for addresses
        containing the recipients full name

      - Collector #953: fixed namespace collision with form_title in ZMI

      - Collector #956: automatically installing Examples at startup
        could be a security risk. Examples now must be installed
        explicitly by the user (thanks to Jamie Heilman and day0).

      - Collector #954: clear() method of TopicIndex removed all filter sets
        instead of clearing them.

      - Collector #882: Fixed typo in PropertySheets

      - Collector #939: Fixed typo in TopicIndexes

      - Collector #937: UnicodeError exception available within PythonScripts

      - Collector #902: recursive Scripts were broken due to shared globals.

      - Product initialization would only consult a file named "version.txt"
        to read version information. Now it will check version.txt, VERSION.txt
        and VERSION.TXT.

      - Make ZCTextIndex much less prone to generating conflict errors.
        Previously *any* concurrent updates would provoke a conflict.

      - Fix query performance and scalability bug in ZCTextIndex.

      - Collector #928: DateIndex ignored timezones when indexing and
        querying

      - Any write request could be tricked into writing into a version
        be setting a version cookie or by including a version name in
        the request. Now we require the user to globally have
        permission to join or leave versions to run a request in a
        version.

      - Fixed a problem with potentially mis-acquiring 'func_code' in
        publisher BeforeTraverse hook.

      - Fix for issue 683: Image cache manager headers were not sent
        when an image request returned a 304 (in response to an if-mod-since
        request).

      - Made all PluginIndexes and ZCTextIndex use 'safe_callable',
        which is aware of extension classes that fill 'tp_callable'
 but don't define '__call__'.

      - Made KeywordIndex be more robust about receiving a value that
        is not a string or an iterable type.

      - Fixed incorrect docstring in OFSP/Image help.

      - Fixed unhelpful signal description.

  Zope 2.6.2 beta 2

    Bugs Fixed

      - TemporaryStorage (which is used by TemporaryFolder, and thus
        the default sessioning configuration) no longer uses a
        "LowConflictConnection" database connection. This fixes
        a bug in which data structures used for session housekeeping
        data could become desynchronized; the symptom for this was
        KeyErrors being raised from TransientObjectContainer's 'get'
        method. As a result, many more conflicts will be raised under
        high session load, but desynchronization will not occur.

      - Fix potential performance bug in PathIndex.

      - Scored result sets from catalog (i.e., from text indexes) can now be
        merged and sorted together across queries like unscored results.

      - Fixed a memory leak in TALES. If an exception propagated from
        a tal:repeat block, an uncollectable cycle held a reference to
        everything in the TALES context.

      - If you're running in development mode, Zope will now
        raise an exception if a product cannot be initialized properly
        instead of silently continuing. This is to prevent a debugging
        frenzy in which you spelunk through one more more half-initialized
        modules wondering why the class you wanted isn't part of the
        module namespace. If not in development mode, the process
        continues silently.

      - Though Python 2.2.2 is not officially supported, a potential
        issue was found when running under 2.2.2. Some built-in types
        gained docstrings in the 2.2.2 release, making them
        publishable where they weren't publishable before. A fix has
        been added to the publisher to ensure that the types of
        objects that are publishable do not change between 2.1 and
        2.2.

  Zope 2.6.2 beta 1

    Bugs Fixed

      - Fixed bug in Missing.Value object that caused a segfault when
        it was compared with certainly builtin types, like unicode.

      - Refactored interface check in ZCatalog indexes view so it is
        no longer broken by the new Zope3 interface logic.

      - Setting multiple selection properties with nothing selected now
        stores an empty list instead of an empty string value.

      - Database invalidations are processed atomically. Each
        transaction will see all the changes caused by an earlier
        transaction or none of them. Before this patch, it was
        possible for a transaction to see invalid data because it saw
        only a subset of the invalidations. This is the most likely
        cause of reported BTrees corruption, where keys were stored in
        the wrong bucket. When a BTree bucket splits, the bucket and
        the bucket's parent are both modified. If a transaction sees
        the invalidation for the bucket but not the parent, the BTree
        in memory will be internally inconsistent and keys can be put
        in the wrong bucket. The atomic invalidation fix prevents
        this problem.

      - Trying to store an object of a non-integer type into an
        IIBTree or OIBTree could leave the bucket in a variety of
        insane states. For example, trying

            b[obj] = "I'm a string, not an integer"

        where b is an OIBTree. This manifested as a refcount leak in
        the test suite, but could have been much worse (most likely in
        real life is that a seemingly arbitrary existing key would "go
        missing").

        When deleting the first child of a BTree node with more than
        one child, a reference to the second child leaked. This could
        cause the entire bucket chain to leak (not be collected as
        garbage despite not being referenced anymore).

        Other minor BTree leak scenarios were also fixed.

      - Collector #683: WeDAV request without XML preamble
        were not recognized by Zope and raised an exception.

      - Fix small bug related to ZCTextIndex attempting to index the
        results of a callable. Do not attempt to index None.

      - Fixed performance degradation when unindexing objects from ZCTextIndex.

      - Improved failure mode of (un)restrictedTraverse to aid debugging.

      - Collector #256 encrypted password bug routed to 2.6 branch

      - Interface package: isImplementedBy() returned None instead of 0.

      - Collector #814: PathIndexes now uses IITreeSet instead of
        IISet. This should result in less memory usage.

      - Collector #816: dtml-sendmail corrupted header if message was
        left blank.

      - Fixed a whitespace problem in Z2.log (AWStats failed to process
        Zope logfiles)

      - It was possible for a transaction that failed in tpc_finish()
        to lose the traceback that caused the failure. The
        transaction code was fixed to report the original error as
        well as any errors that occur while trying to recover from the
        original error.

      - Two small bugs were fixed in DemoStorage. undoLog() did not
        handle its arguments correctly and pack() could accidentally
        delete objects created in versions.

      - Fixed trivial bug in fsrecover that prevented it from working at all.

      - FileStorage will use fsync() on Windows starting with Python 2.2.3.

      - FileStorage's commit version was fixed. It used to stop after
        the first object, leaving all the other objects in the
        version.

      - Collector #892: misleading error msg when initializing an OIBTree
        from a dict with a float value. The message claimed that the
        dict's items didn't consist of 2-element tuples, but of course
        they do. The TypeError now says "expected integer value".

      - Backport of ZODB fix for Colletor #875, where packing to a point
        before an UNDO transaction can lead to objects improperly being
 left out of the packed database (the UNDO maintained a history pointer
 to an object that had been deleted by pack).

  Zope 2.6.1

    Features added

      - DateTime objects now have a tzoffset() method that returns the objects
        timezones offset from GMT in seconds.

      - Fixed bug that caused problems for non-latin1 users who are not
        yet using unicode. plain string properties were always rendered as
        latin-1. The management_page_charset property is now an officially
        accepted migration tool.
        A disadvantage of this change is that it is no longer possible
        to supply an initial value to unicode properties.
        This fixes Collector #737, #598, and #623
        This change was debated here:
        http://lists.zope.org/pipermail/zope-dev/2002-December/018261.html

      - The Transaction "hosed" feature is disabled in this release.
        If a transaction fails during the tpc_finish() it is not
        possible, in general, to know whether the storage is in a
        consistent state. For example, a ZEO server may commit the
        data and then fail before sending confirmation of the commit
        to the client. If multiple storages are involved in a
        transaction, the problem is exacerbated: One storage may
        commit the data while another fails to commit. In previous
        versions of ZODB, the database would set a global variable
        "hosed" that prevented any other transaction from committing
        until an administrator could check the status of the various
        failed storages and ensure that the database is in a
        consistent state. This approach favors data consistency over
        availability. The new approach is to log a panic but
        continue. In practice, availability seems to be more
        important than consistency. The failure mode is exceedingly
        rare in either case.

      - The BTrees-based fsIndex for FileStorage is enabled. This
        version of the index is faster to load and store via pickle
        and uses less memory to store keys. We had intended to enable
        this feature in an earlier release, but failed to actually do
        it; thus, it's getting enabled as a bug fix now.

      - A value-based consistency checker for BTrees was added. See
        the module BTrees.check for the checker and other utilities
        for working with BTrees.

      - The ZEO server has an optional timeout feature that will abort
        a connection that does not commit within a certain amount of
        time. The timeout works by closing the socket the client is
        using, causing both client and server to abort the transaction
        and continue. This is a drastic step, but can be useful to
        prevent a hung client or other bug from blocking a server
        indefinitely.

    Bugs Fixed

      - Collector #721: preserve syntactically valid character entities
        in attributes.

      - I18n interpolation now tries to deal with the case where there
        is a mix of Unicode and non-ascii string that are incompatible
        (because the encoding of the latter is unknown) by substituting
        a representation of the non-ascii string.

      - I18n interpolation doesn't fail anymore if a i18n:name is not
        provided, the ${string} in the translation is just left as is.

      - Collector #696: tal:replace of a non-string (a number for
        example) associated with a i18n:name failed to be interpolated
        properly.

      - If a client was disconnected during a transaction, the
        tpc_abort() call did not properly reset the internal state
        about the transaction. The bug caused the next transaction to
        fail in its tpc_finish().

      - The performance of full cache verification has improved
        dramatically. XXX Get measurements from Jim -- somewhere in
        2x-5x recall. The implementation was fixed to use the
        very-fast getSerial() method on the storage instead of the
        comparatively slow load().

      - Two rare bugs were fixed in BTrees conflict resolution. The
        most probable symptom of the bug would have been a segfault.
        The bugs were found via synthetic stress tests rather than bug
        reports.

      - Transience's '_getCurrentBucket' implementation had a bug which
        could cause more work to be performed than necessary
 (incorrect computation of "plast" variable).

      - Role.py's get_valid_userids method (used by the local roles
        machinery) mistakenly used the name 'acl_users' to locate a user
        folder instead of the internal '__allow_groups__' alias.

      - Collector #771: ZCatalog failed to index DTML Document if the name
        of a catalog metadata was identical with the name of an acquired
        object.

      - Collector #740: DateTime now handles positive numerical
        timezones correcly.

      - Collector #763: There was no error when you had a sendmail-tag
        without specifying a mailhost or smpthost. Also added a
        missing import.

      - Fixed a bug in restrictedTraverse() that sometimes led to an
        AttributeError instead of prompting the user for credentials.

      - Fixed a memory leak in ZODB related to product refresh: when
        the ZODB cache was reset, ZODB still held unnecessary references
        to the old cache.

      - Collector #730: Exceptions are once more logged with their
        tracebacks.

      - Collector #686: intSets no longer loose their values.

      - Collector #685: Improved documentation explaining how, where and 2
        why security assertions should be placed in:
        lib/python/Products/PythonScripts/README.txt
        lib/python/Products/PythonScripts/module_access_examples.py

      - Collector #736: ZPublisher now allows marshalling tags to contain
        a '-'. This is the first step towards a fix for Collector #737

      - Collector #714: CopySupport's manage_clone now calls
        manage_afterClone in the same way that manage_pasteObjects does.

      - Collector #322: made the error message you get when you try to
        use 'client' as a ZSQLMethod argument more meaningful.

      - Collector #479: ZCatalog catalog tab now shows "Unknown" in
        the type column when it doesn't have a meta_type metadata
        value for a cataloged object, rather than the type of the
        catalog itself.

      - Fix bug concatenating Lazy result sets. Also made len potentially
        less expensive for LazyCat objects.

      - Collector #697: Multiple selection properties were incorrectly
        marshalled. note than any non-ascii multiple selection properties
        modified in versions without this fix will have been corrupted in
        the zodb. (this fix was subsequently changed with the fix to #737,
        just before release of 2.6.1 beta 2. Thanks to Maik Jablonski)

      - Collector #256: Added a check in _doChangeUser to make sure
        passwords isn't encrypted twice.

      - Collector #700: History tab of Historical objects caused error
        in storages that did not support history.

      - Collector #595: There was insufficient quoting in some dtml
        files.

      - Collector #561: Ensured that all HTTP headers are sent as normal
        strings and not Unicode (thanks to Lalo Martins).

      - Modify the ZRDB TM class to turn off warnings that Zope DAs do not
        have a sortKey() method.

  Zope 2.6.1 beta 1

    Bugs Fixed

      - VirtualHostMonster handles empty Mapping paths properly.

      - Fixed a bug in ZCTextIndex when trying to raise an error when
        finding the lexicon failed.

      - Major speedups to Catalog sorting code. Sorting will consume less
        memory as well.

      - Collector #597: ZCTextIndex splitters now honor the locale setting
        for proper splitting of non-english text.

      - Deadlock prevention code added.

        It was possible for earlier versions of ZODB to deadlock when
        using multiple storages. If multiple transactions committed
        concurrently and both transactions involved two or more shared
        storages, deadlock was possible. This problem has been fixed
        by introducing a sortKey() method to the transaction and
        storage APIs that is used to define an ordering on transaction
        participants. This solution will prevent deadlocks provided
        that all transaction participants that use locks define a
        valid sortKey() method. A warning is raised if a participant
        does not define sortKey(). For backwards compatibility,
        BaseStorage provides a sortKey() that uses __name__.

      - Fixed bug in FileStorage related to object uncreation. An
        attempt to load an uncreated object now raises KeyError.

      - Fixed a couple bugs in FileStorage recover() that wrote
        incorrect backpointers.

      - Fixed data_txn attribute of iterator data records to use
        the transaction id of the previous transaction, even if it
        also has a data_txn field.

      - Fixed conflict resolution bug that raised a NameError when a
        class involved in a conflict could not be loaded.

      - Fixed C extensions that included standard header files before
        Python.h, which is not allowed.

      - Added code to ThreadedAsync/LoopCallback.py to work around a
        bug in asyncore.py: a handled signal can cause unwanted reads
        to happen.

      - Collector #651: WebDAV Lock Manager was broken.

      - Collector #646: metal:slot was lost during the I18n merge.

      - Collector #640: Fix security assertion on ZCTextIndex query method.

      - Delayed opening the ZODB until after the "Zope" module has
        been imported, fixing a deadlock issue involving ZEO. The
        "Zope" module now has a "startup()" function.

      - Fixed a NameError in the recent change to DateTime.rfc822().

      - Made DateTime.rfc822() simpler and independent of local timezone.

      - Fixed bug in Transience reported by kedai which caused spurious
        KeyErrors under heavy sessioning usage.

      - Fixed bug in the Interface Verify package; base interfaces were not
        included in an interface compliancy test.

      - Collector #650: Fixed implicit list marshalling for lists where the
        first two values are tainted.

      - Collector #671: HTTP Ranges were broken for files and images whose
        length wasn't exactly divisible by 2**16.

      - ModuleSecurityInfo declarations could be lost if further declarations
        were made after the Info object already had been applied. Such
        additional declarations could take place in Python trusted code run
        after Zope strartup or during a Product refresh.

      - Collector #699: MailHosts created in 2.5 breaks in 2.6.

      - Collector #694: dtml-sendmail mailto specification replaces
        "To:" header.

      - Collector #703: KeyErrors raised when unindexing a PathIndex (and
        TopicIndexes) should be swallowed and logged.

  Zope 2.6.0

    Bugs Fixed

      - Caused many places throughout the code base to use
        calls to user.getId() rather than user.getUserName(). With
        most (all?) user folder implementations today, this will have
        no behavioral change, as getId is always alised to getUserName.
        However, this makes it possible to write user folder
        implementations which make the distinction between the user's
        id and the user's name. These user folders will allow users
        to change names independent of their identity.

      - WebDAV Lock Manager actually gives the user a chance to
        specify a starting path **before** searching for locks,
        shortening query times and memory usage in large Zope
        instances.

      - PageTemplateFiles were previously owned by whatever object
        contained them. This resulted in very hard bugs if the user who
        owned the container was removed. Since PageTemplateFiles come
        from the filesystem, they are now "unowned", similar to
        DTMLFiles. Security is still applied, but now it is applied
        correctly.

      - Collector #411: DateTime.rfc822 is not rfc822 compliant

  Zope 2.6.0 beta 2

    Bugs Fixed

      - The ability to add multiple select properties to
        PropertyManagers was broken (issue 612).

      - Removed the signal handler hung off USR1 for packing the database.
        This feature proved dangerous as the pack operation would happen in
        the main thread, causing all asyncore operations to stop until it
        was finished.

      - Collector #372: tal:attributes failed when combined with tal:replace.

      - Don't try to close network connections in the signal handler
        for shutdown. This hosed ZEO clients.

      - Collector #292: PythonScript.write() didn't properly refresh bindings.

      - Dumb bug in zdaemon fixed in which it would try to kill
        process numbers 1, 2, 3, 10, 12, and 15 when it caught a
        signal related to any of these signal numbers. Instead, it
        actually tries now to kill its child process with the same
        signal.

      - Write pidfiles out with trailing newlines.

      - Fix setVirtualRoot in the face of unicode paths (such as occur
        during an XML-RPC request.

      - Collector #539: Fixed rendering of TAL namespace tags with an
        'on-error' statement.

      - Collector #586: Generated 'start' scripts had a nonsensical
        export of an "INST_HOME" environment variable.

      - Collector #580: TALES evaluateBoolean() was squishing 'default'.

      - Collector #581: TALES Path traversal should not special-case a blank
        string in the second element position. It now skips directly
        to item access when a path element is blank or has a leading '_'.

      - Fixed inconsistent attribute access in TALES Paths.

      - Deprecated hasRole alias failed to return result.

      - Collector #538: Hybrid path expressions no longer attempt to call
        a value returned by the final, non-path alternate.

      - Collector #573: ZTUtils Iterator didn't catch AttributeError.

      - Collector #517: The properties page incorrectly rendered properties
        with non-latin1 values if there were no unicode properties defined,
        and incorrectly processed properties with non-ascii names.

      - ZTUtils.SimpleTree could not build a tree with a root other than the
        ZODB root object. Also, filter functions didn't work at all, let
        alone in accordance with the documentation in the code.

      - Collector #603: ZTUtils.Tree.encodeExpansion encoded depth with '.'
        characters, but decodeExpansion could possibly see an encoded node
        id as an encoded depth when that encoded id started with a '.'.

      - Collector #605: ZTUtils.Tree.decodeExpansion set no limits on the
        string to be decoded, allowing for a DoS attack with very large
        strings.

      - The fix for issue #144 broke the ability to create an empty Image or
        File object. This functionality is now reenabled again.

      - ZTUtils.Zope.TreeSkipMixin allows you to skip unauthorized objects in
        the tree, but the filter wasn't applied when trying to filter candidate
        child nodes through a custom setChildAccess filter.

      - Emails sent through MailHost now automatically include a Date header if
        not already present, in compliance with RFC822 and RFC2822.

    Features Added

      - Add optional 'relative' argument to getURL the method in
        CatalogBrains. This allows it to generate site relative URLs
        like absolute_url can.

      - ZTUtils.Tree.encodeExpansion now will use zlib compression by default,
        allowing for a far larger number of open tree states to be encoded.
        decodeExpansion handles compressed expansion states automatically.

      - ZTUtils.Tree.TreeMaker now has additional methods for setting
        various flags and attributes that influence how the tree is built,
        making these aspects accessible to PythonScripts.

      - ZTUtils.Tree.TreeMaker has a new method setStateFunction, which
        allows you to set a callback function that can influence the state
        (open, closed, leaf) of each node in the tree.

      - Pidfile handling improved. When Zope is started under
        zdaemon, it no longer writes its own pidfile. Instead, it
        passes in the path to Z2.pid to zdaemon as its pidfile name.
        The 'zProcessManager.pid' file is no longer ever written.
        This caused a change to the -Z option of z2.py which should be
        mostly backwards-compatible (unless people were relying on
        zProcessManager.pid to be written). Now the -Z option is a
        boolean. -Z1 means use a daemon. -Z0 means dont. The
        default is -Z1.

  Zope 2.6.0 beta 1

    Bugs Fixed

      - Collector #587: fixed wrong migration to string methods in
        DTMLMethod.py

      - Collector #583: Searching for '/' with PathIndexes failed.

      - Fixed bug in manage_editProperties which used an incorrect default
        for several types of property when they were not found in the
        REQUEST.

      - Collector #574: Fixed write on HEAD requests caused by overzealous
        ETag support.

      - Fixed bug in z2.py where it would eat certain socket error exceptions
        at startup.

      - Collector #550: Exceptions in XML-RPC requests no longer envoke
        standard_error_message. Plain text error messages are instead added to
        the fault string. In debug mode, a full traceback is also included
        since access to the error log is not a given for XML-RPC developers.

      - Collector #512,541: Fixed broken WebDAV compatiblity
        with Cadaver 0.20.X due to a missing Lock-Token header.

      - Zope Page Templates set a 'content-type' header even if
        the result of their execution was not rendered to the browser.
        We now check to make sure a content-type header is not
        already set before allowing a page template to set its own.

      - The title_or_id attribute of browser id managers and
        session data managers is now accessible publically.

      - Collector #510: When Python scripts and other "Script" objects were
        acquired during URL traversal, the __before_publishing_traverse__ code
        did not properly stop traversal at the script and populate
        traverse_subpath with the remaining url path elements.

      - Collector #238: Version Save and Discard buttons were too
        close to each other in Version management screens.

      - The "Add Browser ID Manager" permission was renamed to
        "Add Browser Id Manager".

      - Collector #437: dtml-sqltest now renders 'v not in (b,c)'
        when used as <dtml-sqltest v type=... multiple op=ne>.
        Previously, a sqltest for inequality would render 'v <> b'
        when a single value was submitted, but would render
        'a in (b,c)' when multiple values were present and the
        'multiple' switch was set.

      - Collector #478: Z Search Interfaces with no parameters are now
        generating correct HTML.

      - Collector #448: Z Search Interfaces created as PageTemplates
        have a correct title, not a fragment of dtml.

      - Fixed brokenness of session data manager hasSessionData method.
        The old method created a session data object as a result of the
        call; it does not now.

      - Collector #458: Fixed broken reindex_all in CatalogAwareness classes.

      - The default "start" script now causes the event log to be sent to
        standard output unless the "EVENT_LOG_FILE" or "STUPID_LOG_FILE"
        environment variable is found in the environment.

      - The much-hated name "STUPID_LOG_FILE" now has a preferred
        alias: "EVENT_LOG_FILE".

      - Collector #454: The "default" session_data transient object
        container was not created if an object named "session_data"
        existed in the root.

      - Restored behavior of ZCatalog when arguments with empty string are
        passed in to searchResults. These values are now ignored. If only
        empty string values are passed to searchResults, then it returns all
        results (it is assuming what was passed is essentially an empty
        filter).

      - Collector #160: Allow TemporaryStorages to participate
        when a version is active.

      - Collector #446: Fixed management security assertions on
        ZCatalogIndexes class.

      - The BTree module functions weightedIntersection() and
        weightedUnion() now treat negative weights as documented. It's
        hard to explain what their effects were before this fix, as
        the sign bits were getting confused with an internal
        distinction between whether the result should be a set or a
        mapping.

      - New "Transience" (session data storage) implementation.
        More reliable under high load.

      - Collector #402: PythonScript recompile utility should only be
        usable by Manager to prevent abuse.

      - Collector #433: Fixed broken Splitter backwards compatiblity
        issue caused by code cleanup.

      - Collector #151: The Python 2.1 / 2.2 fcntl compatibility hacks
        were bypassed when using medusa directly without importing
        ZServer first (as when using monitor_client.py).

      - Collector #72: Start on Windows 95 machines with no network
        devices installed.

      - Collector #79: Don't swallow App.FindHomes exceptions.

      - The set operation difference(X, None) was returning None
        instead of returning X, contradicting the docs and common
        sense. difference(None, X) continues to return None.

      - Fix bug in ISO_8859_1 splitter which corruped storage on
        initialization.

      - Collector #421: Storage leak in cAccessControl

      - FileLibrary and GuestBook example applications gave anonymous
        users the Manager proxy role when uploading files - a potential
        vulnerability on production servers.

      - Exceptions that use untrusted information from a REQUEST object in
        the exception message now html-quote that information.

      - Stop leaking FastCGI Authorization header in environment to
        prevent password compromise

      - #178: Don't compile PythonScripts in skins directories

      - Fixed the help registration system and Zope tutorial to honor
        the environment variables, FORCE_PRODUCT_LOAD, and ZEO_CACHE,
        that affect whether products are installed in the database at
        application startup.

      - Collector #547: xmlrpclib SlowParser should also handle CDATA
        sections.

      - Collector #525: Don't mask Unautorized exceptions as XML-RPC faults.
        Fix based on patch from Brad Clements.

      - Collector #465: Allow XML-RPC requests with no <params /> tag.

      - Collector #528: Don't clear REQUEST_METHOD for XML-RPC requests;
        instead check for an XML-RPC Response objetc in
        BaseRequest.traverse.

    Features Added

      - Browser ids can now be encoded in the URL and Zope can be
        instructed to automatically include the browser id in its
        generated URLs.

      - Browser Id Managers now provide a saner way to obtain a
        hidden form element which encodes the browser id name and
        browser id. An interface method named "getHiddenFormField"
        on browser id managers now exists which returns a snippet of
        HTML as a hidden form field that encodes these values.

      - A Site Error Log object is now created in the root at Zope
        startup time.

      - Added 'url_unquote' and 'url_unquote_plus' modifiers
        to DTML (also fmt=url-unquote and fmt=url-unquote-plus),
        and made the same functions available in the PythonScripts.standard
        module.

      - Collector #186: Added urlencode to the standard importables for
        Python scripts.

      - <dtml-var name> and &dtml.-name; will now automatically HTML-quote
        unsafe data taken implictly from the REQUEST object. Data taken
        explicitly from the REQUEST object is not affected, as well as any
        other data not originating from REQUEST. This can be disabled (at
        your own risk!) by setting the environment variable
        ZOPE_DTML_REQUEST_AUTOQUOTE to one of 'no', '0', or 'disabled'.

      - ZCatalog index management ui is now integrated into ZCatalog rather
        than being a subobject managment screen with different tabs.

      - ZCTextIndexes can now be instantiated without constructing a silly
        "extra" record object if desired.

      - SimpleItem class now passes a new argument "error_log_url" to
        the standard_error_message template on error. If the site contains
        an error log object, this will contain the url to the applicable log
        entry for the error.

      - The IOBTree module also supports multiunion() now.

      - BTrees and TreeSets are complex objects, with parent->child
        pointers, sibling pointers, and multi-level parent->descendant
        pointers. About half the pointers are formally redundant, but
        speed operations. BTrees and TreeSets now support a ._check()
        method, which does a thorough job of examining all these
        pointers for consistency. It raises AssertionError if it finds
        any problems, else returns None. In Zope 2.5, in rare cases a
        key deletion could leave these internal pointers in an
        inconsistent state (what was supposed to be redundant
        information became conflicting information). The most likely
        symptom was that tree.keys() would yield an object that
        disgreed with the tree about how many keys there are.
        tree._check() can be used if you suspect such a problem (and if
        you find one, rebuilding the tree is the best solution for now).

      - Added support for the ZOPE_HOME environment variable, which
        points to the Zope root, where the ZServer package and default
        imports may be found.

      - Collector #516 -- "title" property on image tags

      - Collector #117 -- change External Method DTML to name="id" vs
        unquoted id

      - Collector #61 -- now manage_PasteObjects return a list of dictionaries
    containing {'id':original_id,'new_id':newly_pasted_obj_id} when called
    with REQUEST=None

      - Changed FORCE_PRODUCT_LOAD so that if it is set, it determines
        whether products are installed regardless of whether ZEO_CACHE is
        set. This means that you can disable product installation by setting
        FORCE_PRODUCT_LOAD to an empty string even if you are not using a
        ZEO persistent cache.

        Documented FORCE_PRODUCT_LOAD

      - xmlrpclib has been updated to the Python 2.2 version, which includes
        support for the Expat parser for unmarshalling data, which speeds up
        things considerably.

      - Binary builds for Linux are now built against glibc 2.1.3 with large
        file support enabled.

      - Binary builds for Solaris are now built against Solaris 8 with large
        file support enabled.

      - Added i18n support in TAL processing

  Zope 2.6.0 alpha 1

    Features Added

      - The IIBTree module has a new multiunion function. It accepts
        a sequence of sets, treesets, etc, and returns the union of
        the keys of these objects, as an IISet. It's designed
        for peak speed when the input sequence contains many objects.

      - Set the default sys checkinterval to a higher value (500) to
        take better advantage of faster processors. Since there is no
        way to scientifically determine a number that works best for
        everyone, this at least should err on the side of better
        performance "out of the box" for higher-end production
        systems.

        Note that you can always use the -i argument to z2 to change
        the check interval.

      - Added support for gzip content compression for clients that
        support it. See lib/python/ZPublisher/HTTPResponse.py for more
        details.

      - Added ZCTextIndex plug-in index product. A replacement for TextIndex.

      - Removed the venerable but senile QuickStart folder from the
        default FileStorage. "Alas, poor Yorick! I knew him, Horatio."

      - Signal handling and log rotation

        All Zope process will respond to signals in the specified manner:

        SIGHUP - close open database connections and sockets, then restart the
                  process

        SIGTERM - close open database connections and sockets, then shut down.

        SIGINT - same as SIGTERM

        SIGUSR2 - rotate all Zope log files (z2.log, event log, detailed log)

        The common idiom for doing automated logfile rotation will become:

         kill -USR2 `cat /path/to/var/z2.pid`

        The common idiom for doing "prophylactic" restarts will become:

         kill -HUP `cat /path/to/var/z2.pid`

        When a process is interrupted via ctrl-C or via a TERM signal
        (INT, TERM), all open database connections and sockets will be closed
        before the process dies. This will speed up restart time for sites
        that use a FileStorage as its index will be written to the filesystem
        before shutdown.

        Unspecified signals kill the process without doing cleanup.

      - ZCatalog no longer has a hand in managing text index vocabularies.
        The cruft associated with this functionality has been exorcised.
        No default indexes or metadata elements are defined for you when
        you create a new ZCatalog. Since we now have many new kinds of
        plug-in indexes it no longer made sense to do this
        anymore.

      - A new permission "Copy or Move" was added. This permission
        may be used respective to an object to prevent objects
        from being copyable or movable while within the management
        interface. The "old" behavior stipulated that users whom
        possessed the "View management screens" permission to an object's
        container could copy or move the object arbitrarily, even if they
        had limited access to the object itself. Once the object was
        moved or copied, the user became the owner of the new object,
        allowing them to see potentially sensitive information in
        the management interface for the object itself. This permission
        is granted to Manager and Anonymous by default, and must be
        revoked on an object-by-object basis if site managers intend
        to provide management screen access to folders which contain
        sensitive subobjects. This patch came as a result of
        Collector #376 (thanks to Chris Deckard).

      - Structured Text's "DocumentWithImages" class did not recognize
        image filenames with underscores.

      - The getElementsByTagName method of STDOM (used by Structured Text)
        would croak on most documents, especially those containing
        unwrapped text nodes. Fixed.

      - FileUpload objects now evaluate false when the have an empty file
        name. Making it easier to check for omitted file upload form fields.

      - ZClasses now use a python script as their constructor method
        instead of a DTML method. Also, ZClasses inherit from
        CatalogPathAwareness now instead of CatalogAwareness.

      - added browser_default hook to ZPublisher. This allows objects to
        specify the path to the default method that the publisher calls
        when the object is published. The default for objects not defining
        browser_default is still 'index_html' for bw compatibility.
        A ZMI configurable browser_default implementation has been added
        to ObjectManager. You can configure browser_default for OMs via
        a new "settings" management tab.

      - added TopicIndexes: a TopicIndex is a container for
        so-called FilteredSet. A FilteredSet consists of an
        expression and a set of internal ZCatalog document
        identifiers that represent a pre-calculated result list for
        performance reasons. Instead of executing the same query on
        a ZCatalog multiple times it is much faster to use a
        TopicIndex instead.

      - requestprofiler: added new --daysago option and added
        support for reading gzipped detailed logfiles

      - DateTime: new functions JulianDay() and week()
        to perform calculation of the week number based on the
        Julian calendar.

      - WebDAV: the new environment variable WEBDAV_SOURCE_PORT_CLIENTS
        enables retrieval of the document source for dedicated WebDAV
        clients (see ENVIRONMENT.txt for usage)

      - Collector #272: Optimizations for RESPONSE.write

      - Collector #271: New environment variables are now used
        to send the access log into syslog. ZSYSLOG_ACCESS,
        ZSYSLOG_ACCESS_FACILITY, and SYSLOG_ACCESS_SERVER now
        do the same job as the old environment variables without
        _ACCESS in their name. Those old environment variables
        still do the same job of sending the event log to syslog.

      - When run as a daemon on Unix, Zope will now redirect
        stdin/stdout/stderr to /dev/null

      - Nicer formatting for the increasingly tall permissions
        table.

      - TextIndex: Enhanced splitter functionality now allows the
        TextIndex to index numbers, single characters. It is also
        possible to enable case-sensitive indexing. The new
        configuration options are available through the addForm
        of the Vocabulary object.

      - ICP server support. For more information see
        http://www.zope.org/Members/htrd/icp/intro

      - STXNG: added new env. variable STX_DEFAULT_LEVEL to change
        the default level for <Hx> elements (see doc/ENVIRONMENT.txt)

      - Collector #304: several catalog optimisations

      - New implementation of ZODB object cache. The new
        implemenation is more likely to keep the size of the object
    cache close to the target size. This change means that memory
    consumption may be reduced. Some users will need to increase
    the default cache size, because a too small setting is more
    likely to hurt performance than it did in the past.

        Third-party C extensions that use the persistence API must be
        recompiled, and may need to be updated to work correctly with
        the new cache; see PER_GHOSTIFY().

      - The ZODB Connection is now resposible for registering changed
        objects with the current transaction.

      - Implementation of RestrictedCreation fishbowl proposal;
        Product registration can now include a function used to
        determine whether that product constructor want to allow
        objects to be created in the specified container object.

      - Collector 196: manage_page_style.css is now cacheable.
        Added freshness information to ImageFile, to improve
        cacheability of management interface

      - Collector 358: added a new parameter no_push_item to
        dtml-in, to inhibit automatically pushing sequence-item
        onto the namespace stack.

      - STXNG: Structured Text now supports images by default
        by using the HTMLWithImages class (has been disabled prior
        to Zope 2.6)

      - new option --force-http-connection-close for z2.py to prevent
        clients from maintaing pipelined connections to the Zope server
        (Collector #412)

      - Updated the Interface package to be compatible with Zope 3
        Interfaces. This included changing some interface APIs that
        may affect existing products.

      - Added a database activity monitoring graph to the control panel,
        making it easier to tune the ZODB cache size.

    Bugs Fixed

      - External methods didn't properly setup func_defaults and func_code
        when they were first loaded. This meant mapply couldn't properly map
        arguments on the first try.

      - Fixed bug #96: Narrower/Wider buttons now work on both CSS and non-CSS
        compliant browsers. This allows better control for browsers that have a
        hard time knowing what 100% means.

      - Fix for Collector #319: filtered_manage_options didn't
        correctly filter tabs based on permission.

      - Made repr of an HTTPRequest.record eval'able as a dict (Collector
        #89).

      - Fixed bug #144: Upload button on dtml, py scripts, images, files and
        pts now raises an error if the file is not specified rather than
        clearing the source.

      - Fixed bug #275: setPermissionDefault didn't actually set the
        right permission -> role mappings.

      - Fixed bug reported on maillist during EWOULDBLOCK when using FTP server
        (http:// lists.zope.org/pipermail/zope/2002-March/111521.html).

      - App/FindHomes.py now computes the "real" path for SOFTWARE_HOME and
        INSTANCE_HOME, resolving any symlinks in any element within paths
        passed in via the INSTANCE_HOME or SOFTWARE_HOME envvars. Paths that
        are computed by "dead reckoning" from os.getcwd and module paths are
        also "realpathed". So for instance, if you use '/home/chrism/Instance'
        as your INSTANCE_HOME, and '/home/chrism' is a symlink to
        '/other/home/chrism', your INSTANCE_HOME will be computed as
        '/other/home/chrism/Instance'. This is necessary to avoid
        weirdnesses while using "dead reckoning" from INSTANCE_HOME and
        SOFTWARE_HOME in other parts of the code. POSIX systems only.

      - Fixed PropertyManager/PropertySheets so that you can safely add a
        property named 'ids' without breaking your properties page.

      - Removed spurious 'self' from scarecrow interfaces; updated
        method-generation in Interface package to ignore self when
        source is a method (rather than a function).

      - Collector #32: Use difflib instead of ndiff

      - Fixed long standing bug in PythonScript where get_size returned
        the incorrect length. This broke editing using EMACS via FTP or
        WebDAV. Thanks to John Glavin at South River Technologies for
        help finding the bug.

      - Collector #207: fixed problem with inner links in STXNG

      - Collector #210: HTML() function of StructuredText produced wrong
        <h0> tags.

      - Collector #166: ObjectManger.all_meta_types() implemented only
        an incomplete filter based on interfaces.

      - FTP: Downloading files through FTP has been broken since 2.4.0
        because the downloaded file has been stored with a HTTP
        header at the beginning of the file. Fixed!

      - FTP: Spaces in usernames inside a FTP file listing are now
        replaced by underscores to avoid confusion with some FTP clients.

      - Collector #227: improved handling of unicode string in TextIndex.py
        with unmodified default encoding in site.py.

      - Collector #227: z2.py, TextIndex/dtml/manage_vocab.dtml modified
        to display unicode strings in the vocabulary properly (now using
        UTF-8 encoding for display purposes)

      - Collector #250: applied several patches for TextIndex for better
        unicode support for the GlobbingLexicon

      - Collector #254: return owner object from getOwner wrapped in its
        context

      - Collector #259: walkandscrub.py did not delete all .pyc and .pyo
        files during installation. Fixed.

      - Collector #231: BTrees ignoring errors from comparison function

      - Collector #278: DocumentWithImages could not handle URLs with
        underscores

      - Collector #279: changed exception handling for safegmtime() to
        provide a more intuitive traceback for operating systems with a
        limited gmtime() implementations

      - Collector #285: Zope changes its working directory
        to the var directory at startup

      - WebDAV: removing an non-existing property now returns a HTTP
        200-OK response instead of 404 (now compliant with RFC 2518)

      - Fixed a bug in TM.py that would cause database adapters to hang
        on errors in the second phase of the two-phase commit.

      - Collector #291: ZCatalog not unindexing deleted properties

      - Collector #266: Retried requests losing track of http request
        headers, causing Connection:Close requests to stall

      - Collector #17: Fixed broken links in StandardCacheManagers help

      - Collector #1: UNIX security fixes: make starting Zope as 'root'
        secure, stop using 'nobody', warn of insecure umasks

      - Collector #303: Properties of type 'long' got truncated

      - Collector #325: adding a new TextIndex to an existing Catalog
        cleared the standard Vocabulary.

      - Collector #373: content_type property for Image objects
        are no longer deletable to prevent malfunction.

      - Collector #343: The ZCatalogs 'Indexes' view showed the
        wrong number of indexed objects for FieldIndexes.

      - FTP server: replaced 'System_Process' by 'Sysproc' to
        avoid breaking some FTP clients and the output format
        with overlong usernames.

      - Fixed a potential bug with cAccessControl's permission
        role deallocator which would try to decref things which
        may not have been set, due to a change in the initializer
        (which will bail out if it doesnt get called with a tuple
        argument)

      - Collector #185, 341: PCGIServer and FCGIServer logs corrected
        and now output extended information like HTTPServer does.

      - Propertysheets: Ids like 'values' and 'items' are
        now forbidden as they break WebDAV functionality. Existing
        Propertysheets are not affected

      - Collector #348: decapitate() now recognizes both \r\n and \n\n
        to be compliant with the HTTP RFC

      - Collector #386: workaround for hanging FTP connections
        with NcFTP

      - Collector #419: repaired off-by-1 errors and IndexErrors
        when slicing BTree-based data structures. For example,
        an_IIBTree.items()[0:0] had length 1 (should be empty) if
        the tree wsan't empty.