- throw out many mentions of "old-style/new-style"
- add memoryview() though I somebody has to fill in the details
- throw out str.decode()
- throw out classobj and instanceobj

exception) and change bytes.find() to use _getbuffer(), so b"".find("")
will raise TypeError instead of SystemError.

with the SDK compiler.  This should fix the failing
Lib\ctypes\test\test_as_parameter.py test.

be UTF-8. Fixes #1070.

line splitting examples don't split things the way they used to -- or should.
In these cases, change the test case and add an XXX.

The final failure was in Charset.body_encode() with euc-jp charset.  These
return the original string unencoded, which isn't right.  XXX and comment this
out for now; we'll fix it after a1.

Fix all remaining instances that did bad things with the new str.translate().

bytes objects don't.

Commit #1068: new docs for PEP 3101. Also document the old string formatting as "old", and begin documenting str/unicode unification.

This fixes test_builtin on windows.

(bug was introduced by the merge of the int/long unification branch,
rev 53421)

mainly concerning PEPs 3131 and 3120.

add -n option to regrtest, and use it on the buildbot.

removed in rev 55543.

(a) Several tests in test_mailbox were failing because we were writing
    text to a file opened in binary mode.  Switching to text fixed these.
(b) test_unix_mbox() in each test does a wacko comparison which apparently
    no longer works due to a different way the message gets parsed.
    I disabled this, I don't think the test was testing what it thought
    it was testing.

2. Re-enable stderr redirection.

Fix a couple of tests since .body_encode()'s arguments have changed.  Also, I
think body_encode() should take a string not a byte array for consistency with
the rest of the api (but I'm not positive about this).  In
quoprimime.body_encode(), body_check() must be passed an int.

Current status: 7F (no errors!)

I had to fix a few markup issues in controlflow.rst and modules.rst.

There's a unicode issue on line 448 in introduction.rst that someone else needs to fix.

MIMEApplication() requires a bytes object for its _data, so fix the tests.

We no longer need utils._identity() or utils._bdecode().  The former isn't
used anywhere AFAICT (where's "make test's" lint? <wink>) and the latter is a
kludge that is eliminated by base64.b64encode().

Current status: 5F/5E

This repairs the linear whitespace insertion between RFC 2047 encoded words
without leaving bogus trailing spaces at the end lines that end in encoded
words.

Current status: 7F/9E