a missing part of the previous checkin message:

Marc-Andre Lemburg:

Added encoding name attributes to wrapper classes which
allow applications to check the used encoding names.

Changed PyUnicode_Splitlines() maxsplit argument to keepends.
The maxsplit functionality was replaced by the keepends
functionality which allows keeping the line end markers together
with the string.

The maxsplit functionality in .splitlines() was replaced by the keepends
functionality which allows keeping the line end markers together
with the string.

Added support for '%r' % obj: this inserts repr(obj) rather
than str(obj).

Added a few missing whitespace Unicode char mappings.
Thanks to Brian Hooper.

The maxsplit functionality in .splitlines() was replaced by the keepends
functionality which allows keeping the line end markers together
with the string.

Added special case to unicode(): when being passed a
Unicode object as first argument, return the object as-is.
Raises an exception when given a Unicode object *and* an
encoding name.

Added .writelines(), .readlines() and .readline() to all
codec classes.

Modified .splitlines() tests according to the changes
in stringobject.c.

Modified .splitlines() tests according to the changes
in unicodeobject.c.

Perhaps these should be added to the standard Python style file?

Lots of scattered wording changes.

IRIX, it doesn't even compile.  Added a cast: "where >= (char *)0".

Add descriptions for the system variables.

(everyone who said something remotely useful in the last 100 messages
I archived has been added :-).

are still C functions, and should be marked.

        Remove statement that the return value is always NULL; this is
        generated by the formatting.

        Fix description; lowercase and uppercase are strings, not
        functions!  Noted by Randall Hopper <aa8vb@yahoo.com>.

maketrans():
        Minor markup nits in description.

PyObject* values.

do_env_cfuncdesc():  Support the "null" value for the refcounts field
        in refcounts.dat.

PyErr_SetFromErrno() should be "null".

reference count field, state that the return value is always NULL.

(Useful for some PyErr_*() flavors.)

        When processing the optional parameter, don't append any text;
        let the pre-processor supply all of it.

Do it in the project file rather than in the source,
so it's easier for Mark to change.

                      of \dataline.

[The test is in a slightly odd place, in test_division_2; but it
exercises the recent change to long_mult(), and that's all we really
ask for.  --GvR]

Kneler for reporting this issue: long_mult() is faster when the
smaller argument is on the left.  Swap the arguments accordingly.

default list of files from () to None, and explicitly test for None
before defaulting to sys.argv[1:].  This means that if you pass in an
explicit empty list, it will read stdin instead of defaulting to
sys.argv[1:].  This fixes a buglet in the test script (when called
with options but without files, it chokes when it tries to interpret
the options as files).

Lawrence adds: "I suspect that this is a safe change, because I can't
imagine someone actively passing in an empty list when they want
sys.argv used."

I agree.

This one should work with unicode expressions, and compile
a bit more silently.

Fredrik Lundh: add two missing casts.

indentation (tabs only), rationalize some code in urljoin...

built-in messages.

allowing it to be run from anywhere, including through a symlink to
the actual idle.py script.