adequate context for the security manager, resulting in excessive Unauthorized
errors.  The previous fix corrected traversal using __bobo_traverse__();
this fix corrects traversal using getattr().  The problem was solved by
simply making use of AccessControl.ZopeGuards.guarded_getattr().

Impacted code:
- TAL: merge of the 2.7 i18n stuff, unicode fixes, tests.
- PageTemplates: addition of a global translation service and of its use
  by the TALES engine, unicode fixes, tests.
- StructuredText: unicode fixes, tests.

merging toby-unicode-size-branch, a fix for a regression introduced by TaintedString changes, reported as collector #565

wide xterm or emacs window.  So now, when using Python 2.3 or later,
the textwrap module is used to wrap the message line, indenting
subsequent lines past the timestamp.  Details and tracebacks are left
alone.

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

access in certain situations when using a __bobo_traverse__ hook.

__bobo_traverse__ hook, and the hook returns an object that comes from
some other container, the security policy may incorrectly deny access.
This fix determines the container of the object based on its
aquisition wrappers, if available, and passes it to validate().

Also added a corresponding unit test.

should fix the problem that messages from different processes or
threads could be intermingled.  The code was rewritten to build up the
lines of the error message in a list which is joined by "\n"
characters at the end.

- Used a MappingStorage instead of a FileStorage, to avoid writing to disk.

- Removed unnecessary docstrings and converted useful docstrings to comments
to clean up the unit test output.

- Updated to use unittest.main().

throw either AttributeError or KeyError; catch either.

loadmail() when deciding on the print format.

Unicode-aware StringIO by monkey-patching two appropriate methods
until one is completely integrated into Zope.

- You can pass the list of indexes to update to catalog_object().

- uids can be relative rather than absolute.

lexicon.  Before this, it was not safe to move, rename, or mount in a
different location a ZCatalog containing a ZCTextIndex.

In general, we need to avoid physical paths when a simple ID will suffice.

response.
(Same change as ChrisM's in ZopePageTemplate.py)

directory.

any suitable tests produced an ugly traceback. Instead the
testrunner now prints a warning to stderr.

Added an argument to pass error_log_url into raise_standardErrorMessage since logging is now done above it.

After further investigation I found that my change to SimpleItem to sniff the response._error_format was unnecessary (although harmless) since Zope/__init__.py already does this. This check has been removed.

And since error logging was being done in SimpleItem, my previous change prevented xml-rpc exceptions from being logged.

The logging code has now been moved into the Zope/__init__ top-level exception handler to correct this problem. This also seems generally more correct since we no longer rely on SimpleItem to do-the-right-thing.

Refactored error message handling so that SimpleItem's raise_standardErrorMessage sniffs the response to see if it wants an html encoded error message or not.

If so, then standard_error_message is rendered as usual as the error value. Otherwise the error value text is passed through unchanged.

XML-RPC responses now request text/plain error format so that standard_error_message is not envoked to render the fault string. Because of this, the previous formatting code could be removed, although html tags are still stripped because error values themselves sometimes contain tags 8^/

 - Value is stripped of HTML tags and minimally formatted

 - In debug mode, a full traceback is included in the fault string

occurs.  First, inherit from BerkeleyTestBase which has proper setUp
and tearDown methods, and second, wrap the .abort() and ._close() call
in tearDown() in a try/finally so we're guaranteed that the base
class's tearDown gets run.

likely to actually remove the test-db directory every time.

for 'index_html' default views. This breaks things like Cookie Crumbler. And
yes, when you use a JavaScript XML-RPC library, the browser could include an
authentication cookie for CookieCrumbler.

Instead, do not clear REQUEST_METHOD but check for a xmlrpc.Response object
when initiating the variable that will cause 'index_html' views to be used.

This fixes Zope Collector issue #528.