pickle.py
The code implicitly assumed that all ints fit in 4 bytes, causing all
sorts of mischief (from nonsense results to corrupted pickles).
Repaired that.
marshal.c
The int marshaling code assumed that right shifts of signed longs
sign-extend. Repaired that.
of 2-space and 4-space indents. Whatever, when I saw the checkin diff it
was clear that what my editor thinks a tab means didn't match this module's
belief. Removed all the tabs from the lines I added and changed, left
everything else alone.
pickled into the signed(!) 4-byte BININT format, so were getting unpickled
again as negative ints. Repaired that.
Added some minimal docs at the top about what I've learned about the pickle
format codes (little of which was obvious from staring at the code,
although that's partly because all the size-related bugs greatly obscured
the true intent of the code).
Happy side effect: because save_int() needed to grow a *proper* range
check in order to fix this bug, it can now use the more-efficient BININT1,
BININT2 and BININT formats when the long's value is small enough to fit
in a signed 4-byte int (before this, on a sizeof(long)==8 box it always
used the general INT format for negative ints).
test_cpickle works again on sizeof(long)==8 machines. test_pickle is
still busted big-time.
bugs on sizeof(long)==8 machines. pickle.py has no idea what it's
doing with very large ints, and variously gets things right by accident,
computes nonsense, or generates corrupt pickles. cPickle fails on
cases 2**31 <= i < 2**32: since it *thinks* those are 4-byte ints
(the "high 4 bytes" are all zeroes), it stores them in the (signed!) BININT
format, so they get unpickled as negative values.
(Yes, this is a new feature right before the 2.1 release. No, I can't
imagine this would seriously break anybody's code. In fact, most
users of this script are probably *happy* to see this addition.)
binary pickle, and the latter contains a pickle of a negative Python
int i written on a sizeof(long)==4 box (and whether by cPickle or
pickle.py), it's read incorrectly as i + 2**32. The patch repairs that,
and allows test_cpickle.py (to which I added a relevant test case earlier
today) to work again on sizeof(long)==8 boxes.
There's another (at least one) sizeof(long)==8 binary pickle bug, but in
pickle.py instead. That bug is still there, and test_pickle.py doesn't
catch it yet (try pickling and unpickling, e.g., 1 << 46).
think that a command starting with '#' is a comment, so move the one
comment in such a position (in the rule for building $(LIBRARY)) to a
harmless position.
integers, but the std tests don't exercise most of them. Repair that.
CAUTION: I expect this to fail on boxes with sizeof(long)==8, in the
part of test_cpickle (but not test_pickle) trying to do a binary mode
(not text mode) load of the embedded BINDATA pickle string. Once that
hypothesized failure is confirmed, I'll fix cPickle.c.
Even though relative redirects are illegal, they are common
urllib treated every relative redirect as though it was to http,
even if the original was https://
As long as we're compensating for server bugs, might as well do
it properly.
Add mangling support
Add get_children() and add_child() methods to Scope
Skip nodes when If test is a false constant
Add test code that checks results against symtable module
Fix com_NEWLINE() so that is accepts arguments, which occurs for lines like:
stmt; # note trailing semicolon
Add XXX about checking for assignment to list comps
where sizeof(long)==8. This *was* broken on boxes where signed right
shifts didn't sign-extend, but not elsewhere. Unfortunately, apart
from the Cray T3E I don't know of such a box, and Guido has so far
refused to buy me any Cray machines for home Python testing <wink>.
More immediately interesting would be if someone could please test
this on *any* sizeof(long)==8 box, to make sure I didn't break it.
http://groups.yahoo.com/group/medusa/message/333
It's clear that Medusa should not be checking for an empty buffer
via "buf is ''". The patch merely changes "is" to "==". However,
there's a mystery here all the same: Python attempts to store null
strings uniquely, so it's unclear why "buf is ''" ever returned
false when buf actually was empty. *Some* string operations produce
non-unique null strings, e.g.
>>> "abc"*0 is "abc"*0
0
>>>
but they're rare, and I don't see any such operations in asynchat.
Jeffery Collins pointed out that filterstring decrefs a character object
before it's done using it. This works by accident today because another
module always happens to have an active reference too at the time. The
accident doesn't work after his Pippy modifications, and since it *is*
an accident even in the mainline Python, it should work by design there too.
The patch accomplishes that.
Steve Purcell's documentation, and a lot of it is written based on
using PyUnit and reading the implementation.
There is more to come, but I want to get this check in before I have a
disk crash or anything else bad happens.
add a self-test using doctest. Results:
- The docstring needs to be a raw string because it uses \"...\".
- The oreo example was broken: the Set-Cookie output doesn't add
quotes around "doublestuff".
- I had to change the example that prints the class of a Cookie.Cookie
instance to avoid incorporating an arbitrary object address in the
test output.
Pretty good score for both doctest and the doc string, I'd say!
Tim Peters