interned when created, so the cached versions generally aren't ever
interned.  With the patch, the
		Py_INCREF(t);
		*p = t;
		Py_DECREF(s);
		return;
indirection block in PyString_InternInPlace() is never executed during a
full run of the test suite, but was executed very many times before.  So
I'm trading more work when creating one-character strings for doing less
work later.  Note that the "more work" here can happen at most 256 times
per program run, so it's trivial.  The same reasoning accounts for the
patch's simplification of string_item (the new version can call
PyString_FromStringAndSize() no more than 256 times per run, so there's
no point to inlining that stuff -- if we were serious about saving time
here, we'd pre-initialize the characters vector so that no runtime testing
at all was needed!).

Store floats and doubles to full precision in marshal.
Test that floats read from .pyc/.pyo closely match those read from .py.
Declare PyFloat_AsString() in floatobject header file.
Add new PyFloat_AsReprString() API function.
Document the functions declared in floatobject.h.

calldll is part of the standard macPython distribution.

Documentation update to reflect changes to the termios module (noting
that the termios functions can take a file object as well as a file
descriptor).

This closes the documentation portion of SF patch #417081.

d1 == d2 and d1 != d2 now work even if the keys and values in d1 and d2
don't support comparisons other than ==, and testing dicts for equality
is faster now (especially when inequality obtains).

Assertion error message had typos in arguments to string format.

.cover files for modules in packages are now put in the right place.

The code that generate .cover files seemed to prepend a "./" to many
absolute paths, causing them to fail.  The code now checks explicitly
for absolute paths and leaves them alone.

In trace/coverage code, recover from case where module has no __name__
attribute, when e.g. it is executed by PyRun_String().  In this case,
assign modulename to None and hope for the best.  There isn't anywhere
to write out coverage data for this code anyway.

Also, replace several sys.stderr.writes with print >> sys.stderr.

New features:

-C/--coverdir dir: Generate .cover files in specified directory
instead of in the directory where the .py file is.

-s: Print a short summary of files coverred (# lines, % coverage,
name)

Check for free in class and method only if nested scopes are enabled.

Add assertion to verify that no free variables occur when nested
scopes are disabled.

XXX When should nested scopes by made non-optional on the trunk?

Two errors in dict_to_map() helper used by PyFrame_LocalsToFast().

another change (to test_import.py, which simply imports the new file).  I'm
checking this piece in now, though, to make it easier to distribute a patch
for x-platform checking.

2.1.1 bugfix candidate too.
Fix a bad (albeit unlikely) return value in try_rich_to_3way_compare().
Also document do_cmp()'s return values.

This patch does several things to termios:

(1) changes all functions to be METH_VARARGS
(2) changes all functions to be able to take a file object as the
    first parameter, as per

http://mail.python.org/pipermail/python-dev/2001-February/012701.html

(3) give better error messages
(4) removes a bunch of comments that just repeat the docstrings
(5) #includes <termio.h> before #including <sys/ioctl.h> so more
    #constants are actually #defined.
(6) a couple of docstring tweaks

I have tested this minimally (i.e. it builds, and
doesn't blow up too embarassingly) on OSF1/alpha and
on one of the sf compile farm's solaris boxes, and
rather more comprehansively on my linux/x86 box.

It still needs to be tested on all the other platforms
we build termios on.


This closes the code portion of SF patch #417081.

Wrapped a long line.

NEEDS DOC CHANGES.
More AttributeErrors transmuted into TypeErrors, in test_b2.py, and,
again, this strikes me as a good thing.
This checkin completes the iterator generalization work that obviously
needed to be done.  Can anyone think of others that should be changed?

safely together and don't duplicate logic (the common logic was factored
out into new private API function _PySequence_IterContains()).
Visible change:
    some_complex_number  in  some_instance
no longer blows up if some_instance has __getitem__ but neither
__contains__ nor __iter__.  test_iter changed to ensure that remains true.

   but don't plan on actually writing any text until, ooh, say, July or
   thereabouts.

NEEDS DOC CHANGES
A few more AttributeErrors turned into TypeErrors, but in test_contains
this time.
The full story for instance objects is pretty much unexplainable, because
instance_contains() tries its own flavor of iteration-based containment
testing first, and PySequence_Contains doesn't get a chance at it unless
instance_contains() blows up.  A consequence is that
    some_complex_number in some_instance
dies with a TypeError unless some_instance.__class__ defines __iter__ but
does not define __getitem__.

to string.join(), so that when the latter figures out in midstream that
it really needs unicode.join() instead, unicode.join() can actually get
all the sequence elements (i.e., there's no guarantee that the sequence
passed to string.join() can be iterated over *again* by unicode.join(),
so string.join() must not pass on the original sequence object anymore).

because PySequence_Fast() started working for free as soon as
PySequence_Tuple() learned how to work with iterators.  For some reason
unicode.join() still doesn't work, though.

several of these turned up and got fixed during the iteration crusade.

NEEDS DOC CHANGES.
This one surprised me!  While I expected tuple() to be a no-brainer, turns
out it's actually dripping with consequences:
1. It will *allow* the popular PySequence_Fast() to work with any iterable
   object (code for that not yet checked in, but should be trivial).
2. It caused two std tests to fail.  This because some places used
   PyTuple_Sequence() (the C spelling of tuple()) as an indirect way to test
   whether something *is* a sequence.  But tuple() code only looked for the
   existence of sq->item to determine that, and e.g. an instance passed
   that test whether or not it supported the other operations tuple()
   needed (e.g., __len__).  So some things the tests *expected* to fail
   with an AttributeError now fail with a TypeError instead.  This looks
   like an improvement to me; e.g., test_coercion used to produce 559
   TypeErrors and 2 AttributeErrors, and now they're all TypeErrors.  The
   error details are more informative too, because the places calling this
   were *looking* for TypeErrors in order to replace the generic tuple()
   "not a sequence" msg with their own more specific text, and
   AttributeErrors snuck by that.

internals) so clients can be a lot dumber (wrt their knowledge).

free to do one!

NEEDS DOC CHANGES.

NEEDS DOC CHANGES.
Possibly contentious:  The first time s.next() yields StopIteration (for
a given map argument s) is the last time map() *tries* s.next().  That
is, if other sequence args are longer, s will never again contribute
anything but None values to the result, even if trying s.next() again
could yield another result.  This is the same behavior map() used to have
wrt IndexError, so it's the only way to be wholly backward-compatible.
I'm not a fan of letting StopIteration mean "try again later" anyway.

at the same time as it did from PyObject_Init() .

the code necessary to accomplish this is simpler and faster if confined to
the object implementations, so we only do this there.

This causes no behaviorial changes beyond a (very slight) speedup.

void function.

need to be specified in the type structures independently.  The flag
exists only for binary compatibility.

This is a "source cleanliness" issue and introduces no behavioral changes.

NEEDS DOC CHANGES.

    When replacing the exception object, be sure we stuff the new value
    in sys.last_value (which we already did for the original value).