changes.  Here restoring them.

libm result is 0).  Cautiously add a few libm exception test cases:
1. That exp(-huge) returns 0 without exception.
2. That exp(+huge) triggers OverflowError.
3. That sqrt(-1) raises ValueError specifically (apparently under glibc linked
   with -lieee, it was raising OverflowError due to an accident of the way
   mathmodule.c's CHECK() macro happened to deal with Infs and NaNs under gcc).

and it would look kind of silly if typeset.

   left that wasn't covered.  The article is therefore now essentially complete.)
A few minor changes

assuming here that the ANSI-C adjacent-string-concatenation technique is
allowable, now that Python requires an ANSI C compiler.

Py_FatalError() instead, and clarify the message somewhat. As discussed on
python-dev.

Moved appendChild calls from DOMEventStream to PullDOM (parser indep).
Removed duplicated sibling pointer setting (duplicated in appendChild).

support for extension types, not classes.

pickle():  If the type is a class or if the reduction function is not
           callable, raise a TypeError.

constructor():  If the constructor is not callable, raise TypeError.

This (partially) closes SourceForge patch #101859.

MPZ_divm():  Initialize mpzden to NULL, since it could be Py_XDECREF()ed
             without being initialized.

string.letters was much more than expected.

Solution: explicit is better than implicit; don't rely on
string.letters.

a bug in JPython where the instance had to have a flush() method.

groupings are still maintained.  Based on a comment by Steve Holden
<sholden@holdenweb.com>.

on Win64.

This closes bug http://sourceforge.net/bugs/?func=detailbug&group_id=5470&bug_id=116516

variable.  w should be initialized before entering the bytecode
interpretation loop since we only need one initialization to satisfy the
compiler.

by making the DUP_TOPX code utterly straightforward.  This also gets rid
of all normal-case internal DUP_TOPX if/branches, and allows replacing one
POP() with TOP() in each case, so is a good idea regardless.

Based on a comment from Detlef Lannert
<lannert@lannert.rz.uni-duesseldorf.de>.

<lannert@lannert.rz.uni-duesseldorf.de>.

these computations are required for their side effects in traversing the
variable arguments list.

Reported by Marc-Andre Lemburg <mal@lemburg.com>.

than the built-in string types (suggested by Moshe Zadka
<moshez@math.huji.ac.il>).

Clarified what "can be converted to a string" means.

Fixed a few markup nits.

one comment in the example for clarity.

Yee <ping@lfw.org> and probably others as well.

file-system accessible repository.  Add a little bit of smarts to convert
the cvsroot to an anonymous cvsroot the real one requires an authenticated
login to SourceForge; this avoids the SSH startup delay when doing the
checkout or export to get a fresh copy of the tree.

only for extension types.

This partially fixes SourceForge bug #116295.

Added "See also" section with relevant RFC references.

and queue-management modules, since that is where the general context
for use in most cases.

Who know where the handle will point to tomorrow?

didn't bother to close the files.  This caused the new test_wave test to fail
under Windows, as Windows won't let you delete a file that's open.  Fixed
that by ensuring the wave read & write classes' .close() and __del__ methods
close files that were opened by their constructors.