change in meaning.

of my recent XXX comment, taking a (what appears to be vanishingly small)
chance and calling self.memoize() instead.

marshal.loads() beats struct.unpack()!  Possibly because the latter
creates a one-tuple. :-(

40% faster than marshal.dumps(x)[1:]!  (That's not counting the
module attribute lookups, which can be avoided in either case.)

created by recursive tuples.

iteritems() instead of materializing a (possibly giant) list of the
items.

The code is much easier to follow now, and I bet it's faster too.

make the bin-vs-not-bin order consistent with what other routines try to
do (they almost all handle the bin case first).

some notion of low-level efficiency.  Undid that, but left one routine
alone:  save_inst() claims it has a reason for not using memoize().
I don't understand that comment, so added an XXX comment there.

belongs.  This is a much smaller change than it may appear:  the bulk
of the function merely got unindented by one level.

then the embedded argument consumes at least 256 bytes.  The difference
between a 3-byte prefix (LONG2 + 2 bytes) and a 5-byte prefix (LONG4 +
4 bytes) is at worst less than 1%.  Note that binary strings and binary
Unicode strings also have only "size is 1 byte, or size is 4 bytes?"
flavors, and I expect for the same reason.  The only place a 2-byte
thingie was used was in BININT2, where the 2 bytes make up the *entire*
embedded argument (and now EXT2 also does this); that's a large savings
over 4 bytes, because the total opcode+argument size is so small in
the BININT2/EXT2 case.

Removed the TAKEN_FROM_ARGUMENT "number of bytes" code, and bifurcated it
into TAKEN_FROM_ARGUMENT1 and TAKEN_FROM_ARGUMENT4.  Now there's enough
info in ArgumentDescriptor objects to deduce the # of bytes consumed by
each opcode.

Rearranged the order in which proto2 opcodes are listed in pickle.py.

The protocol now defaults to 1.  Protocol 2 is still unimplemented.

(SF patch #675750, to fix SF bug #675552.)

dict from the Unpickler's memo (which is a different beast!).

expensive way to spell chr() and ord().

component strings by a blank instead of a period.  Guido pointed
out that the component strings (at least the first one) can be
dotted already.  find_class() is overridable too, so only God knows
all the possibilities that make sense to someone.

this over the weekend, and it made faster & better progress than I
expected -- it's already useful <wink>.