I only ran it as a script.

Added computed attribute for mapply to make up for removal of func_* attributes from normal state in previous fix

was really stupid. It caused problems because func_defaults might
be non-unpicklable (e.g. includes somemodule.rfind).

Added simple unit test to allow me to fix the bug.

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

to be compliant with the HTTP RFC

signature.

1. Out by one error where it would not unravel links in the final path element
2. Correct handling of link destinations which are relative paths
This needs a unit test, but doesnt yet have one.

I *think* this fixes the misleading test failure on Windows Python 2.1.*.

Committing patch from Albert Ting on Zope maillist 4/12.  Regex change which fixes a table parsing problem.

from ExtensionClass, and reminded me that the 2.1 version of isinstance()
does not deal with the more general constructs that newer versions of
Python can handle.

or absolute column widths. You can set this in the standard prefs. Still only
works for editing DTML right now.

    .CatalogAware) as a
    ZClass base class.  Its reason for existence is to make the name
    that shows up in the ZClass Product list different than 'ZCatalog:
    CatalogAware', which is the name registered by
    CatalogAwareness.CatalogAware.  The fix should *really* be to
    change the product registry to keep the whole module/class path
    and to make the ZClass add UI show the whole path, but this is
    nontrivial, we don't want to spend a lot of time on ZClasses, and
    this works.

    Now when we tell people to use CatalogPathAware instead of CatalogAware,
    they will actually be able to do it without a lot of effort.

    The reason we can't replace CatalogAware with CatalogPathAware is
    backwards-compatibility and the desire to not force people to do
    data conversion on their existing instances.

in affect for DTML thus far.