patch #617312, both on the trunk and the 22-maint branch.

Also added a test case, and ported the test_trace I wrote for HEAD
to 2.2.2 (with all those horrible extra 'line' events ;-).

CCompiler may rely on the presence of self.verbose (SciPy's distutils
appears to).

they'll end up in the right directory when --dir is used.

needs to be expanded by mkhowto, since the shell won't touch it except
at the start of the complete argument.

default when used for 3rd-party docs.  This requires the standard Python docs
to specify a location, but they are a bit of a special case in sharing one set
of icons among several documents.

Barry, please add a reasonable abstract!

from an attributes description.

\verbatiminput instead of the verbatim environment -- this does the "right
thing" regarding page breaks in long examples for the typeset formats, and
has nice benefits for the HTML version as well.

I'll do next).

Strangely, two out of three patches there seem already committed; but
the essential one (get rid of the assert in object_filenames in
ccompiler.py) was not yet applied.

This makes the build procedure for Twisted work again.

This is *not* a backport candidate despite the fact that identical
code appears to exist in 2.2.2; Twisted builds fine there, so there
must have been a change elsewhere.

loadTestsFromName to accept new-style classes too!

I seriously doubt this will break any deployed code.

Docstring consistency with the updated .tex files.

semantics of header chunks using byte and Unicode strings.
Specifically,

append(): When the given string is a byte string, charset (whether
specified explicitly in the argument list or implicitly via the
constructor default) is the encoding of the byte string, and a
UnicodeError will be raised if the string cannot be decoded with that
charset.  If s is a Unicode string, then charset is a hint specifying
the character set of the characters in the string.  In this case, when
producing an RFC 2822 compliant header using RFC 2047 rules, the
Unicode string will be encoded using the following charsets in order:
us-ascii, the charset hint, utf-8.

__init__(): Use the global USASCII Charset instance when the charset
argument is None.  Also, clarification in the docstring.

Also, use True/False where appropriate.

Relax regular expression to handle spaces in user and group names.

    ``void (*)(PyObject *)``

isn't quite the same as

    void (*)(PyObject \*)

but I gather ``...`` is preferable stylistically in most cases than
... \*...

Python 2.1.3.  However it's required by the email tests suite, so poke
it into the encodings aliases if it's missing.  The is apparently the
approved API for doing so.

Now we can remove the hexversion shortcircuits in the test suite.