Highlights:
- Adding PyObject_Format.
- Adding string.Format class.
- Adding __format__ for str, unicode, int, long, float, datetime.
- Adding builtin format.
- Adding ''.format and u''.format.
- str/unicode fixups for formatters.
The files in Objects/stringlib that implement PEP 3101 (stringdefs.h,
unicodedefs.h, formatter.h, string_format.h) are identical in trunk
and py3k. Any changes from here on should be made to trunk, and
changes will propogate to py3k).
They still remain fragile.
For example, a call to assertEqual currently does not make any allocation
(which surprised me at first).
But this can change when gc.collect also deletes the numerous "zombie frames"
attached to each function.
./python.exe -m timeit -s 'from fractions import Fraction; f = Fraction(3, 2)' 'isinstance(3, Fraction); isinstance(f, Fraction)'
from 12.3 usec/loop to 3.44 usec/loop and
./python.exe -m timeit -s 'from fractions import Fraction' 'Fraction(3, 2)'
from 48.8 usec to 23.6 usec by avoiding genexps and sets in __instancecheck__
and inlining the common case from __subclasscheck__.
to GET, so it loses its payload. So, it also must lose the
headers related to the payload (if it has no content any more,
it shouldn't indicate content length and type).
* Convert both to unittest.
* Use the same unit testing code.
* Include the expected output in both test files.
* Make it possible to regenerate the expected output by running
the file as a script with an '-r' argument.
The mapping between bytecode offsets and source lines (lnotab) did not contain
an entry for the beginning of the loop.
Now it does, and the lnotab can be a bit larger:
in particular, several statements on the same line generate several entries.
However, this does not bother the settrace function, which will trigger only
one 'line' event.
The lnotab seems to be exactly the same as with python2.4.
I implemented the function sys._compact_freelists() and C API functions PyInt_/PyFloat_CompactFreeList() to compact the pre-allocated blocks of ints and floats. They allow the user to reduce the memory usage of a Python process that deals with lots of numbers.
The patch also renames sys._cleartypecache to sys._clear_type_cache
which isn't as bad as it sounds. The close() *should* raise an exception,
so if it didn't we should give more time to sync and really raise it.
Will backport.
global to determine when the server is ready to be used. This slows the test
down, but should make it correct. There was a race condition before where the
server could have assigned a port, yet it wasn't ready to serve requests. If
the client sent a request before the server was completely ready, it would get
an exception. There was machinery to try to handle this condition. All of
that should be unnecessary and removed if this change works. A NOTE was
added as a comment about what needs to be fixed.
The buildbots will tell us if there are more errors or
if this test is now stable.
in case there were transient failures. This will hopefully silence
the buildbots for this test. As we find other tests that have a problem,
we can fix with a similar strategy assuming it is successful. It worked
on my box in a loop for 10+ runs where it would have an exception otherwise.
whole construct away, even when an 'else' clause is present::
while 0:
print("no")
else:
print("yes")
did not generate any code at all.
Now the compiler emits the 'else' block, like it already does for 'if' statements.
Will backport.
type of encrypted zip files. Files using extended local headers
needed to compare the check byte against different values. (according
to reading the infozip unzip crypt.c source code)
Fixes issue1003.
precision. This has been discussed at http://bugs.python.org/issue1682. It's
useful primarily for teaching, but it also demonstrates how to implement a
member of the numeric tower, including fallbacks for mixed-mode arithmetic.
I expect to write a couple more patches in this area:
* Rational.from_decimal()
* Rational.trim/approximate() (maybe with different names)
* Maybe remove the parentheses from Rational.__str__()
* Maybe rename one of the Rational classes
* Maybe make Rational('3/2') work.
Surprising behaviour of the "$" regexp: it matches the
end of the string, AND just before the newline at the end
of the string::
re.sub('$', '#', 'foo\n') == 'foo#\n#'
Python is consistent with Perl and the pcre library, so
we just document it.
Guido prefers "\Z" to match only the end of the string.
round included:
* Revert round to its 2.6 behavior (half away from 0).
* Because round, floor, and ceil always return float again, it's no
longer necessary to have them delegate to __xxx___, so I've ripped
that out of their implementations and the Real ABC. This also helps
in implementing types that work in both 2.6 and 3.0: you return int
from the __xxx__ methods, and let it get enabled by the version
upgrade.
* Make pow(-1, .5) raise a ValueError again.
* Make the _replace() method respect subclassing.
* Using property() to make _fields read-only wasn't a good idea.
It caused len(Point._fields) to fail.
* Add note to _cast() about length checking and alternative with the star-operator.
the complex_pow part), r56649, r56652, r56715, r57296, r57302, r57359, r57361,
r57372, r57738, r57739, r58017, r58039, r58040, and r59390, and new
documentation. The only significant difference is that round(x) returns a float
to preserve backward-compatibility. See http://bugs.python.org/issue1689.
r58207 and r58247 patch logic is reversed. I noticed this when I
tried to use urllib to retrieve a file which required auth.
Fix that and add a test for 401 error to verify.
Also fix the test by having the test classes inherit from object.
Are the getter/setter/deleter attributes supposed to be able to chain? As of
right now they can't as the property tries to call what the property returns,
which is another property when they are chained.
This changes the rules for when __hash__ is inherited slightly,
by allowing it to be inherited when one or more of __lt__, __le__,
__gt__, __ge__ are overridden, as long as __eq__ and __ne__ aren't.
This adds signal.set_wakeup_fd(fd) which sets a file descriptor to
which a zero byte will be written whenever a C exception handler runs.
I added a simple C API as well, PySignal_SetWakeupFd(fd).
Issue #1580: New free format floating point representation based on "Floating-Point Printer Sample Code", by Robert G. Burger. For example repr(11./5) now returns '2.2' instead of '2.2000000000000002'.
Thanks to noam for the patch! I had to modify doubledigits.c slightly to support X64 and IA64 machines on Windows. I also added the new file to the three project files.
on 32-bit systems on 64-bit systems, and vice versa. As a consequence
of the change, Random pickles created by Python 2.6 cannot be loaded
in Python 2.5.
Added PyFloat_GetMax(), PyFloat_GetMin() and PyFloat_GetInfo() to the float API.
Added a dictionary sys.float_info with information about the internal floating point type to the sys module.
(in deallocation of running threads, for example), so the PyGILState_Release()
function must still be functional.
On the other hand, _PyGILState_Fini() only frees memory, and can be called later.
Backport candidate, but only after some experts comment on it.
number: now it does not store the mantissa as a tuple
of numbers, but as a string.
This avoids a lot of conversions, and achieves a
speedup of 40%. The API remains intact.
Thanks Mark Dickinson.
newlines; it doesn't. To rectify this the string returned replaces all
instances of os.linesep with '\n' to fake universal newline support.
Backport candidate.
as usual with slicing (both with str and unicode strings). This
fixes issue 1259.
For str only the stringobject.c file was modified. But for unicode,
I needed to repeat in the four functions a lot of code, so created
a new function that does part of the job for them (and placed it in
find.h, following a suggestion of Barry).
Also added tests for this behaviour.
(suprisingly, this simplifies the signature, improves clarity, and is comparably fast).
Update the docs to reflect a previous change to the function name.
Add an example to the docs showing how to override the default __repr__ method.
Correction for issue1265 (pdb bug with "with" statement).
When an unfinished generator-iterator is garbage collected, PyEval_EvalFrameEx
is called with a GeneratorExit exception set. This leads to funny results
if the sys.settrace function itself makes use of generators.
A visible effect is that the settrace function is reset to None.
Another is that the eventual "finally" block of the generator is not called.
It is necessary to save/restore the exception around the call to the trace
function.
This happens a lot with py3k: isinstance() of an ABCMeta instance runs
def __instancecheck__(cls, instance):
"""Override for isinstance(instance, cls)."""
return any(cls.__subclasscheck__(c)
for c in {instance.__class__, type(instance)})
which lets an opened generator expression each time it returns True.
Backport candidate, even if the case is less frequent in 2.5.
coefficient numbers, floats in the sign, and other details that
generated directly the wrong number in the best case, or triggered
misfunctionality in the alorithms.
Test cases added for these issues. Thanks Mark Dickinson.
it closes itself. When the stream is read in several calls to read(n),
it should behave in the same way if HTTPConnection knows where the end
of the stream is (through self.length). Added a test case for this
behaviour.
This should make the tests more robust at the expense of allowing
tests to be sloppier by not requiring them to cleanup after themselves.
(It will legitamitely help when running two test suites simultaneously
or if another process is already using one of the predefined ports.)
Also simplifies (slightLy) the exception handling elsewhere.
Let the field spec be either a string or a non-string sequence (suggested by Martin Blais with use cases).
Improve the error message in the case of a SyntaxError (caused by a duplicate field name).
Add maxlen support to deque() and fixup docs.
Partially fix __reduce__(). The None as a third arg was no longer supported.
Still needs work on __reduce__() to handle recursive inputs.
with Cawlishaw by mail, and he basically confirmed that to these is_*
operations, there's no need to return Decimal(0) and Decimal(1) if
the language supports the False and True booleans.
Also added a few tests for the these functions in extra.decTest, since
they are mostly untested (apart from the doctests).
Thanks Mark Dickinson
predictable to being completely predictable. The value of hash(n)
is unchanged for any n that's small enough to be representable as an
int, and also unchanged for the vast majority of long integers n of
reasonable size.
Allows optional commas in the field-name spec (help when named tuples are used in conjuction with sql queries).
Adds the __fields__ attribute for introspection and to support conversion to dictionary form.
Adds a __replace__() method similar to str.replace() but using a named field as a target.
Clean-up spelling and presentation in doc-strings.
adding the 'makefile' method to ssl.SSLSocket, and importing the
requisite fakefile class from socket.py, and making the appropriate
changes to it to make it use the SSL connection.
Added sample HTTPS server to test_ssl.py, and test that uses it.
Change SSL tests to use https://svn.python.org/, instead of
www.sf.net and pop.gmail.com.
Added utility function to ssl module, get_server_certificate,
to wrap up the several things to be done to pull a certificate
from a remote server.
fully updated to the latests Decimal Specification (v1.66) and the
latests test cases (v2.56).
Thanks to Mark Dickinson for all his help during this process.
Backport abc.py and isinstance/issubclass overloading to 2.6.
I had to backport test_typechecks.py myself, and make one small change
to abc.py to avoid duplicate work when x.__class__ and type(x) are the
same.
Make sure the type of the return value of re.sub(x, y, z) is the type
of y+x (i.e. unicode if either is unicode, str if they are both str)
even if there are no substitutions or if x==z (which triggered various
special cases in join_list()).
Could be backported to 2.5; no need to port to 3.0.
* Much expanded test suite:
All protocols tested against all other protocols.
All protocols tested with all certificate options.
Tests for bad key and bad cert.
Test of STARTTLS functionality.
Test of RAND_* functions.
* Fixes for threading/malloc bug.
* Issue 1065 fixed:
sslsocket class renamed to SSLSocket.
sslerror class renamed to SSLError.
Function "wrap_socket" now used to wrap an existing socket.
* Issue 1583946 finally fixed:
Support for subjectAltName added.
Subject name now returned as proper DN list of RDNs.
* SSLError exported from socket as "sslerror".
* RAND_* functions properly exported from ssl.py.
* Documentation improved:
Example of how to create a self-signed certificate.
Better indexing.
alone class. This addresses the primary concern in
http://bugs.python.org/issue1706815
python-dev discussion here:
http://mail.python.org/pipermail/python-dev/2007-July/073749.html
I chose IOError rather than EnvironmentError as the base class since
socket objects are often used as transparent duck typed file objects
in code already prepared to deal with IOError exceptions.
also a minor fix:
urllib2 - fix a couple places where IOError was raised rather than URLError.
for better or worse, URLError already inherits from IOError so
this won't break any existing code.
test_urllib2net - replace bad ftp urls.
ever going back out to Python code in PyObject_Call(). Required introducing a
static RuntimeError instance so that normalizing an exception there is no
reliance on a recursive call that would put the exception system over the
recursion check itself.
1) Improve the documentation of the SSL module, with a fuller
explanation of certificate usage, another reference, proper
formatting of this and that.
2) Fix Windows bug in ssl.py, and general bug in sslsocket.close().
Remove some unused code from ssl.py. Allow accept() to be called on
sslsocket sockets.
3) Use try-except-else in import of ssl in socket.py. Deprecate use of
socket.ssl().
4) Remove use of socket.ssl() in every library module, except for
test_socket_ssl.py and test_ssl.py.
- Specialcase extended slices that amount to a shallow copy the same way as
is done for simple slices, in the tuple, string and unicode case.
- Specialcase step-1 extended slices to optimize the common case for all
involved types.
- For lists, allow extended slice assignment of differing lengths as long
as the step is 1. (Previously, 'l[:2:1] = []' failed even though
'l[:2] = []' and 'l[:2:None] = []' do not.)
- Implement extended slicing for buffer, array, structseq, mmap and
UserString.UserString.
- Implement slice-object support (but not non-step-1 slice assignment) for
UserString.MutableString.
- Add tests for all new functionality.
> the better option. Then the openssl command in the test code can be turned
> into a comment describing how the test data was pregenerated.
Here's a patch that does that.
Bill
exceptions raised in the test server thread, since SimpleXMLRPCServer
does not gracefully handle them. Changed number of requests handled
by tests server thread to one (was 2) because no tests require more
than one request. [GSoC - Alan McIntyre]
1) Fixes the bug that two class names are initial-lower-case.
2) Replaces the poll waiting for the server to become ready with
a threading.Event signal.
object's co_consts tuple; add a test to show that the previous behavior (where
these two constants were "collapsed" into one) causes serious malfunctioning.
and tests for NOOP, RSET, and VRFY. Corrected typo in a comment for
testNonnumericPort. Added a check for constructing SMTP objects when
non-numeric ports are included in the host name. Derived a server from
SMTPServer to test various ESMTP/SMTP capabilities. Check that a
second HELO to DebuggingServer returns an error. [GSoC - Alan McIntyre]
that allows sending back exception/stack trace information about
internal server errors (this flag defaults to False to avoid sending
such information unless explicitly enabled). Added tests to verify
behavior of this new feature (these tests are skipped on win32 because
of problems with WSAEWOULDBLOCK). Renamed HTTPTestCase to
SimpleServerTestCase. [GSoC - Alan McIntyre]
against a SimpleXMLRPCServer in a separate thread. Because of
problems with WSAEWOULDBLOCK (error 10035) being raised by the server
on Windows, these new tests are skipped on win32. [GSoC - Alan McIntyre]
test.test_support.catch_warning is more full-featured and provides the same
functionality.
Since guard_warnings_filter was added in 2.6 there is no
backwards-compatibility issues.
DebuggingServerTests, construct SMTP objects with a localhost argument
to avoid abysmally long FQDN lookups (not relevant to items under
test) on some machines that would cause the test to fail. Moved server
setup code in the server function inside the try block to avoid the
possibility of setup failure hanging the test. Minor edits to conform
to PEP 8. [GSoC - Alan McIntyre]
overrides asyncore.dispatcher.handle_expt to do nothing, since
select.poll gives a POLLHUP error at the completion of these tests.
Added timeout & count arguments to several asyncore.loop calls to
avoid the possibility of a test hanging up a build. [GSoC - Alan
McIntyre]
find_prefix_at_end function. Check behavior of a string given as a
producer. Added tests for behavior of asynchat.async_chat when given
int, long, and None terminator arguments. Added usepoll attribute to
TestAsynchat to allow running the asynchat tests with poll support
chosen whether it's available or not (improves coverage of asyncore
code). [GSoC - Alan McIntyre]
so that the event gets set if a failure occurs during server setup
(otherwise the test will block forever). Changed to let the OS assign
the server port number, and client side of test waits for port number
assignment before proceeding. The test data in DispatcherWithSendTests
is also sent in multiple send() calls instead of one to make sure this
works properly. [GSoC - Alan McIntyre]
smtpd.DebuggingServer. Change to use global host & port number
variables. Modified the 'server' to take a string to send back in
order to vary test server responses. Added a test for the reaction of
smtplib.SMTP to a non-200 HELO response. [GSoC - Alan McIntyre]
to server & client, and by adding asyncore.close_all calls in
tearDown. Also choose correct expected logging results based on the
value of __debug__ [Alan McIntyre - GSoC]
The Maildir specification doesn't seem to say anything about this
situation, and it can happen if you're keeping a Maildir mailbox in
Subversion (.svn directories) or some similar system. The patch just
ignores directories in the cur/, new/, tmp/ folders.
a large width is passed on 32-bit platforms. Found by Google.
It would be good for people to review this especially carefully and verify
I don't have an off by one error and there is no other way to cause overflow.
with tests in test_urllib2net.py (must have network resource
enabled to execute them). Also modified test_urllib2.py because
testing mock classes must take it into acount. Docs are also
updated.
specify an error handling scheme for character conversion. Additional
scheme "utf-8" in read mode. Unicode input filenames are now
supported by design. The values of the pax_headers dictionary are now
limited to unicode objects.
Fixed: The prefix field is no longer used in PAX_FORMAT (in
conformance with POSIX).
Fixed: In read mode use a possible pax header size field.
Fixed: Strip trailing slashes from pax header name values.
Fixed: Give values in user-specified pax_headers precedence when
writing.
Added unicode tests. Added pax/regtype4 member to testtar.tar all
possible number fields in a pax header.
Added two chapters to the documentation about the different formats
tarfile.py supports and how unicode issues are handled.
Python 2.5.
Also remove gopher support from urllib/urllib2. As both imported gopherlib the
usage of the support would have raised a DeprecationWarning.
are zero. This should help reduce the false positives.
The message about references leaking is maintained to provide as much
info as possible rather than simply suppressing the message at the source.
If the call to requires() doesn't precede the filesystem check, we get the following situation:
1. ./python Lib/test/regrtest.py test_foo # test needs urlfetch, not enabled, so skipped
2. ./python Lib/test/regrtest.py -u urlfetch test_foo # test runs
3. ./python Lib/test/regrtest.py test_foo # test runs (!)
By moving the call to requires() *before* the filesystem check, the fact that fetched files are cached on the local disk becomes an implementation detail, rather than a semantics-changing point of note.
http://mail.python.org/pipermail/python-dev/2007-March/071796.html .
I've kept a couple of still-valid extra tests in test_descr, but didn't
bother to sort through the new comments and refactorings added in r53997
to see if some of them could be kept. If so, they could go in a
follow-up check-in.
Add code to read from master_fd in the parent, breaking when we get an OSError
(EIO can occur on Linux) or there's no more data to read. Without this,
test_pty.py can hang on the waitpid() because the child is blocking on the
stdout write. This will definitely happen on Mac OS X and could potentially
happen on other platforms. See the comment for details.
locally. Now, it checks if have openssl available and run
those specific tests (it starts openssl at the beggining of
all the tests and then kills it at the end).
locale changed but not used during the function call it was recreated during.
The test in this checkin is untested (OS X does not have the proper locale
support for me to test), although the fix for the bug this deals with
was tested by the OP (#1290505). Once the buildbots verify the test at least
doesn't fail it becomes a backport candidate.
changes in the tests, so failing one test won't produce strange
results in others. Also relaxed the timeout settings in the test
(where actually the value didn't mean anything).
to complex using its __complex__() method before falling back to the
__float__() method. Therefore, the functions in the cmath module now
can operate on objects that define a __complex__() method.
(backport)
with ASCII value less than 32. Also, it correctly quotes dots only if
they occur on a single line, as opposed to the previous behavior of
quoting dots if they are the second character of any line.
support.
The TarInfo class now contains all necessary logic to process and
create tar header data which has been moved there from the TarFile
class. The fromtarfile() method was added. The new path and linkpath
properties are aliases for the name and linkname attributes in
correspondence to the pax naming scheme.
The TarFile constructor and classmethods now accept a number of
keyword arguments which could only be set as attributes before (e.g.
dereference, ignore_zeros). The encoding and pax_headers arguments
were added for pax support. There is a new tarinfo keyword argument
that allows using subclassed TarInfo objects in TarFile.
The boolean TarFile.posix attribute is deprecated, because now three
tar formats are supported. Instead, the desired format for writing is
specified using the constants USTAR_FORMAT, GNU_FORMAT and PAX_FORMAT
as the format keyword argument. This change affects TarInfo.tobuf()
as well.
The test suite has been heavily reorganized and partially rewritten.
A new testtar.tar was added that contains sample data in many formats
from 4 different tar programs.
Some bugs and quirks that also have been fixed:
Directory names do no longer have a trailing slash in TarInfo.name or
TarFile.getnames().
Adding the same file twice does not create a hardlink file member.
The TarFile constructor does no longer need a name argument.
The TarFile._mode attribute was renamed to mode and contains either
'r', 'w' or 'a'.
nests test.test_support.TransientResource context managers that capture
exceptions raised when the Internet connection is flaky.
Initially using in test_socket_ssl but should probably be expanded to cover any
test that should not raise the captured exceptions if the Internet connection
works.
Patch #1591665: implement the __dir__() special function lookup in PyObject_Dir.
Had to change a few bits of the patch because classobjs and __methods__ are still
in Py2.6.
surround calls to resources that may or may not be available. Specifying the
expected exception and attributes to be raised if the resource is not available
prevents overly broad catching of exceptions.
This is meant to help suppress spurious failures by raising
test.test_support.ResourceDenied if the exception matches. It would probably
be good to go through the various network tests and surround the calls to catch
connection timeouts (as done with test_socket_ssl in this commit).
We add some new rules that are required for preserving internal
invariants of types.
1. If type (or a subclass of type) appears in bases, it must appear
before any non-type bases. If a non-type base (like a regular
new-style class) occurred first, it could trick type into
allocating the new class an __dict__ which must be impossible.
2. There are several checks that are made of bases when creating a
type. Those checks are now repeated when assigning to __bases__.
We also add the restriction that assignment to __bases__ may not
change the metaclass of the type.
Add new tests for these cases and for a few other oddball errors that
were no previously tested. Remove a crasher test that was fixed.
Also some internal refactoring: Extract the code to find the most
derived metaclass of a type and its bases. It is now needed in two
places. Rewrite the TypeError checks in test_descr to use doctest.
The tests now clearly show what exception they expect to see.
Fixes bug 1569356, but at the cost of a minor incompatibility in
locals(). Add test that verifies that the class namespace is not
polluted. Also clarify the behavior in the library docs.
Along the way, cleaned up the dict_to_map and map_to_dict
implementations and added some comments that explain what they do.
The next step of PEP 352 (for 2.6) causes raising a string exception to trigger
a TypeError. Trying to catch a string exception raises a DeprecationWarning.
References to string exceptions has been removed from the docs since they are
now just an error.
generic so that one only has to shift certain values based on whether the week
was specified to start on Monday or Sunday. Cut out a lot of edge case code
compared to the previous version. Also broke algorithm out into its own
function (that is private to the module).
Fixes bug #1643943 (thanks Biran Nahas for the report).
doesn't support the same funcationality as on Unix. I'm not sure if
this fix is the best (or if it will even work)--it's a test to see
if the buildbots start passing again.
It might be better to not even run this test if it's windows (or non-posix).
the master should close the slave fd.
Added a test to test_pty.py that reads from the master_fd after doing
a pty.fork(); without the fix it hangs forever instead of raising an
exception. (<crossing fingers for the buildbots>)
2.5 backport candidate.
described, and add a test for it.
2.5 bugfix candidate, maybe; arguably this patch changes the API of
dumbdbm and shouldn't be added in a point-release.
This change looks massive but it's mostly a re-indenting after
removing some try...finally blocks.
Also adds a test case that does a pack() while the mailbox is locked; this
test would have turned up bugs in the original code on some platforms.
In both nmh and GNU Mailutils' implementation of MH-format mailboxes,
no locking is done of individual message files when renaming them.
The original mailbox.py code did do locking, which meant that message
files had to be opened. This code was buggy on certain platforms
(found through reading the code); there were code paths that closed
the file object and then called _unlock_file() on it.
Will backport to 25-maint once I see how the buildbots react to this patch.
It seems like this should be a different error than SystemError, but
I don't have any great ideas and SystemError was raised in 2.4 and earlier.
Will backport.
* unified the way intobject, longobject and mystrtoul handle
values around -sys.maxint-1.
* in general, trying to entierely avoid overflows in any computation
involving signed ints or longs is extremely involved. Fixed a few
simple cases where a compiler might be too clever (but that's all
guesswork).
* more overflow checks against bad data in marshal.c.
* 2.5 specific: fixed a number of places that were still confusing int
and Py_ssize_t. Some of them could potentially have caused
"real-world" breakage.
* list.pop(x): fixing overflow issues on x was messy. I just reverted
to PyArg_ParseTuple("n"), which does the right thing. (An obscure
test was trying to give a Decimal to list.pop()... doesn't make
sense any more IMHO)
* trying to write a few tests...
The compiler was checking that there was something on the fblock
stack, but not that there was a loop on the stack. Fixed that and
added a test for the specific syntax error.
Bug fix candidate.
As mentioned on python-dev, reverting patch #1504333 because it introduced
an infinite loop in rev 47154.
This patch also adds a test to prevent the regression.
- gbk and gb18030 codec now handle U+30FB KATAKANA MIDDLE DOT correctly.
- iso2022_jp_2 codec now encodes into G0 for KS X 1001, GB2312
codepoints to conform the standard.
- iso2022_jp_3 and iso2022_jp_2004 codec can encode JIS X 2013:2
codepoints now.
generator expressions (x for x, in ... ) works again.
Sigh, I only fixed for loops the first time, not list comps and genexprs too.
I couldn't find any more unpacking cases where there is a similar bug lurking.
This code should be refactored to eliminate the duplication. I'm sure
the listcomp/genexpr code can be refactored. I'm not sure if the for loop
can re-use any of the same code though.
Will backport to 2.5 (the only place it matters).
Debian sparc buildbots. Since this goes through a lot of tests
and hits the disk a lot it could be slow (especially if NFS is involved).
I'm not sure if that's the problem, but printing periodic msgs shouldn't hurt.
The code was stolen from test_compiler.
OverflowError while x*x succeeds and produces infinity; apparently
these inconsistencies cannot be fixed across ``all'' platforms and
there's a widespread feeling that therefore ``every'' platform
should keep suffering forevermore. Ah well.
Small: Always generate a NL or NEWLINE token following
a COMMENT token. The old code did not generate an NL token if
the comment was on a line by itself.
Large: The output of untokenize() will now match the
input exactly if it is passed the full token sequence. The
old, crufty output is still generated if a limited input
sequence is provided, where limited means that it does not
include position information for tokens.
Remaining bug: There is no CONTINUATION token (\) so there is no way
for untokenize() to handle such code.
Also, expanded the number of doctests in hopes of eventually removing
the old-style tests that compare against a golden file.
Bug fix candidate for Python 2.5.1. (Sigh.)
inf) but didn't; added a test to test_float to verify that, and ignored the
ERANGE value for errno in the pow operation to make the new test pass (with
help from Marilyn Davis at the Google Python Sprint -- thanks!).
sporadically on other platforms. This is really a band-aid that doesn't
fix the underlying issue in SocketServer. It's not clear if it's worth
it to fix SocketServer, however, I opened a bug to track it:
http://python.org/sf/1540386
Replace UnicodeDecodeErrors raised during == and !=
compares of Unicode and other objects with a new
UnicodeWarning.
All other comparisons continue to raise exceptions.
Exceptions other than UnicodeDecodeErrors are also left
untouched.
were failing due to inappropriate clipping of numbers larger than 2**31
with new-style classes. (typeobject.c) In reviewing the code for classic
classes, there were 2 problems. Any negative value return could be returned.
Always return -1 if there was an error. Also make the checks similar
with the new-style classes. I believe this is correct for 32 and 64 bit
boxes, including Windows64.
Add a test of classic classes too.
I modified this patch some by fixing style, some error checking, and adding
XXX comments. This patch requires review and some changes are to be expected.
I'm checking in now to get the greatest possible review and establish a
baseline for moving forward. I don't want this to hold up release if possible.
protected by "if verbose:", which caused the test to fail on
all non-Windows boxes.
Note that I deliberately didn't convert this to unittest yet,
because I expect it would be even harder to debug this on Tru64
after conversion.
appears to be utterly insane. Plug some theoretical
insecurities in the test script:
- Verify that the SIGALRM handler was actually installed.
- Don't call alarm() before the handler is installed.
- Move everything that can fail inside the try/finally,
so the test cleans up after itself more often.
- Try sending all the expected signals in
force_test_exit(), not just SIGALRM. Since that was
fixed to actually send SIGALRM (instead of invisibly
dying with an AttributeError), we've seen that sending
SIGALRM alone does not stop this from hanging.
- Move the "kill the child" business into the finally
clause, so the child doesn't survive test failure
to send SIGALRM to other tests later (there are also
baffling SIGALRM-related failures in test_socket).
- Cancel the alarm in the finally clause -- if the
test dies early, we again don't want SIGALRM showing
up to confuse a later test.
Alas, this still relies on timing luck wrt the spawned
script that sends the test signals, but it's hard to see
how waiting for seconds can so often be so unlucky.
test_threadedsignals: curiously, this test never fails
on Tru64, but doesn't normally signal SIGALRM. Anyway,
fixed an obvious (but probably inconsequential) logic
error.
The first hunk changes the colon to an ! like other Windows variants.
We need to always wait on the child so the lock gets released and
no other tests fail. This is the try/finally in the second hunk.
at stopping test_signal from hanging forever on the Tru64
buildbot. That could be because there's no such thing as
signal.SIGALARM. Changed to the idiotic (but standard)
signal.SIGALRM instead, and added some more debug output.
64-bit boxes. I have no idea what the ctypes docs mean
by "integers", and blind-guessing here that it intended to
mean the signed C "int" type, in which case perhaps I can
repair this by feeding the thread id argument to type
ctypes.c_long().
Also made the worker thread daemonic, so it doesn't hang
Python shutdown if the test continues to fail.
of quoted test data relied on preserving a single trailing
blank. Changed the string from raw to regular, and forced
in the trailing blank via an explicit \x20 escape.
PyThreadState_SetAsyncExc(): internal correctness changes wrt
refcount safety and deadlock avoidance. Also added a basic test
case (relying on ctypes) and repaired the docs.
directories each time it ran, at least on Windows.
Several changes: explicitly closed all files; wrapped long
lines; stopped suppressing errors when removing a file or
directory fails (removing /shouldn't/ fail!); and changed
what appeared to be incorrect usage of os.removedirs() (that
doesn't remove empty directories at and /under/ the given
path, instead it must be given an empty leaf directory and
then deletes empty directories moving /up/ the path -- could
be that the conceptually simpler shutil.rmtree() was really
actually intended here).
In the 2.5 development cycle, MAKE_CLOSURE as changed to take free
variables as a tuple rather than as individual items on the stack.
Closes patch #1534084.
on each iteration. I'm not positive this is the best way to handle
this. I'm also not sure that there aren't other cases where
the lnotab is generated incorrectly. It would be great if people
that use pdb or tracing could test heavily.
Also:
* Remove dead/duplicated code that wasn't used/necessary
because we already handled the docstring prior to entering the loop.
* add some debugging code into the compiler (#if 0'd out).
Patch by Douglas Greiman.
The test_run_abort() testcase produces a core file on Unix systems,
even though the test is successful. This can be confusing or alarming
to someone who runs 'make test' and then finds that the Python
interpreter apparently crashed.
These problems may mask more important, real problems.
One or both methods are known to fail on: Solaris, OpenBSD, Debian, Ubuntu.
They pass on Windows and some Linux boxes.
41667, 41668 - initial switch to xmlcore
47044 - mention of xmlcore in What's New
50687 - mention of xmlcore in the library reference
re-apply xmlcore changes to xml:
41674 - line ending changes (re-applied manually), directory props
41677 - add cElementTree wrapper
41678 - PSF licensing for etree
41812 - whitespace normalization
42724 - fix svn:eol-style settings
43681, 43682 - remove Python version-compatibility cruft from minidom
46773 - fix encoding of \r\n\t in attr values in saxutils
47269 - added XMLParser alias for cElementTree compatibility
additional tests were added in Lib/test/test_sax.py that failed with
the xmlcore changes; these relate to SF bugs #1511497, #1513611
concept, and that different ways of trying to find "the
hardware address" may return different results. Certainly
true on both of my Windows boxes, and in different ways
(see whining on python-dev).
inspect.py, and pydoc.py. Specifically, this allows for querying the type of
an object against these built-in C types and more importantly, for getting
their docstrings printed in the interactive interpreter's help() function.
This patch includes a new built-in module called _types which provides
definitions of getset and member descriptors for use by the types.py module.
These types are exposed as types.GetSetDescriptorType and
types.MemberDescriptorType. Query functions are provided as
inspect.isgetsetdescriptor() and inspect.ismemberdescriptor(). The
implementations of these are robust enough to work with Python implementations
other than CPython, which may not have these fundamental types.
The patch also includes documentation and test suite updates.
I commit these changes now under these guiding principles:
1. Silence is assent. The release manager has not said "no", and of the few
people that cared enough to respond to the thread, the worst vote was "0".
2. It's easier to ask for forgiveness than permission.
3. It's so dang easy to revert stuff in svn, that you could view this as a
forcing function. :)
Windows build patches will follow.
When a node number disagrees, keep track of all sources & the
node numbers they reported, and stick all that in the error message.
Changed all callers to supply a non-empty "source" argument; made
the "source" argument non-optional.
On my box, test_uuid still fails, but with the less confusing output:
AssertionError: different sources disagree on node:
from source 'getnode1', node was 00038a000015
from source 'getnode2', node was 00038a000015
from source 'ipconfig', node was 001111b2b7bf
Only the last one appears to be correct; e.g.,
C:\Code\python\PCbuild>getmac
Physical Address Transport Name
=================== ==========================================================
00-11-11-B2-B7-BF \Device\Tcpip_{190FB163-5AFD-4483-86A1-2FE16AC61FF1}
62-A1-AC-6C-FD-BE \Device\Tcpip_{8F77DF5A-EA3D-4F1D-975E-D472CEE6438A}
E2-1F-01-C6-5D-88 \Device\Tcpip_{CD18F76B-2EF3-409F-9B8A-6481EE70A1E4}
I can't find anything on my box with MAC 00-03-8a-00-00-15, and am
not clear on where that comes from.
to guess /which/ line the spawned thread is in at the time
sys._current_frames() is called: we know it finished
enter_g.set(), but can't know whether the instruction
counter has advanced to the following leave_g.wait().
The latter is overwhelming most likely, but not guaranteed,
and I see that the "x86 Ubuntu dapper (icc) trunk" buildbot
found it on the other line once. Changed the test so it
passes in either case.
that a 2.93 sec audio file will always take 3.1 sec (as it did on the
hardware I had when I first wrote the test), expect that it will take
2.93 sec +/- 10%, and only fail if it's outside of that range.
Compute the expected
Moved the code for _PyThread_CurrentFrames() up, so it's no longer
in a huge "#ifdef WITH_THREAD" block (I didn't realize it /was/ in
one).
Changed test_sys's test_current_frames() so it passes with or without
thread supported compiled in.
Note that test_sys fails when Python is compiled without threads,
but for an unrelated reason (the old test_exit() fails with an
indirect ImportError on the `thread` module). There are also
other unrelated compilation failures without threads, in extension
modules (like ctypes); at least the core compiles again.
Do we really support --without-threads? If so, there are several
problems remaining.
of values in the time tuple passed in. Unfortunately people came to rely on
undocumented behaviour of setting unneeded values to 0, regardless of if it was
within the valid range. Now those values force the value internally to the
minimum value when 0 is passed in.
The hppa ubuntu box sometimes hangs forever in these tests. My guess
is that the wait is failing for some reason. Use WNOHANG, so we won't
wait until the buildbot kills the test suite.
I haven't been able to reproduce the failure, so I'm not sure if
this will help or not. Hopefully, this change will cause the test
to fail, rather than hang. That will be better since we will get
the rest of the test results. It may also help us debug the real problem.
*** The reason this originally failed was because there were many
zombie children outstanding before rev 47158 cleaned them up.
There are still hangs in test_subprocess that need to be addressed,
but that will take more work. This should close some holes.
str() or repr()) would work, just not multi-value tuples. Probably not a
backport candidate, since it changes the behaviour of passing a
single-element tuple:
>>> string.Template("$foo").substitute(dict(foo=(1,)))
'(1,)'
versus
'1'
handler would cause a segfault. This merges in Expat's lib/xmlparse.c
revisions 1.154 and 1.155, which fix this and a closely related problem
(the later does not affect Python).
Moved the crasher test to the tests for xml.parsers.expat.
be called at the end of each test that spawns children (perhaps it
should be called from regrtest instead?). This will hopefully prevent
some of the unexplained failures in the buildbots (hppa and alpha)
during tests that spawn children. The problems were not reproducible.
There were many zombies that remained at the end of several tests.
In the worst case, this shouldn't cause any more problems,
though it may not help either. Time will tell.
(modified patch by Sam Ruby; changed to use separate REs for start and end
tags to reduce matching cost for end tags; extended tests; updated to avoid
breaking previous changes to support IPv6 addresses in unquoted attribute
values)
consistent with os.wait() returning immediately because some other
subprocess had previously exited; the test suite then immediately
tries to lock the mailbox and gets an error saying it's already
locked.
To fix this, do a waitpid() so the test suite only continues once
the intended child process has exited.
('[' and ']' were not accepted in unquoted attribute values)
- cleaned up tests of character and entity reference decoding so the
tests cover the documented relationships among handle_charref,
handle_entityref, convert_charref, convert_codepoint, and
convert_entityref, without bringing up Unicode issues that sgmllib
cannot be involved in
both a subclass of Warning and a subclass of types.ClassType. The latter is no
longer true thanks to new-style exceptions.
Closes bug #1510580. Thanks to AMK for the test.
The problem of checking too eagerly for recursive calls is the
following: if a RuntimeError is caused by recursion, and if code needs
to normalize it immediately (as in the 2nd test), then
PyErr_NormalizeException() needs a call to the RuntimeError class to
instantiate it, and this hits the recursion limit again... causing
PyErr_NormalizeException() to never finish.
Moved this particular recursion check to slot_tp_call(), which is not
involved in instantiating built-in exceptions.
Backport candidate.
constructor, meaning it is treated as *args, not as a single argument. This
means using the 'message' attribute won't work (until Py3K comes around),
and so one must grab from 'arg' to get the error number.
os.environ (setting envar COLUMNS), which at least caused
test_float_default() to fail if the tests were run more than once.
This repairs the test_optparse -R failures Neal reported on
python-dev. It also explains some seemingly bizarre test_optparse
failures we saw a couple weeks ago on the buildbots, when
test_optparse failed due to test_file failing to clean up after
itself, and then test_optparse failed in an entirely different
way when regrtest's -w option ran test_optparse a second time.
It's now obvious that make_parser() permanently changing os.environ
was responsible for the second half of that.
is that the wait is failing for some reason. Use WNOHANG, so we won't
wait until the buildbot kills the test suite.
I haven't been able to reproduce the failure, so I'm not sure if
this will help or not. Hopefully, this change will cause the test
to fail, rather than hang. That will be better since we will get
the rest of the test results. It may also help us debug the real problem.
issues that crop up from time to time, but this change seems to have been
pretty stable (no spurious warnings) for about a week.
Other modules which use threads may require similar use of
threading_setup/threading_cleanup from test_support.
of this test. It probably still requires more disk space
than most buildbots have, and in any case is still so
intrusive that if we don't find another way to test this I'm
taking my buildbot offline permanently ;-)
This will hopefully get the buildbots to pass. Not sure this
test will be feasible or even work. But everything is red now,
so it can't get much worse.
arguments in reverse, the interpreter would infinitely recourse trying to get a
coercion that worked. So put in a recursion check after a coercion is made and
the next call to attempt to use the coerced values.
Fixes bug #992017 and closes crashers/coerce.py .
Heavily revised, comprising revisions:
46640 - original trunk revision (backed out in r46655)
46647 - markup fix (backed out in r46655)
46692:46918 merged from branch aimacintyre-sf1454481
branch tested on buildbots (Windows buildbots had problems
not related to these changes).
when running with -O.
test_simple_validation_error still fails under -O. That appears to be because
wsgiref's validate.py uses `assert` statements all over the place to check
arguments for sanity. That should all be changed (it's not a logical error
in the software if a user passes bogus arguments, so this isn't a reasonable
use for `assert` -- checking external preconditions should generally raise
ValueError or TypeError instead, as appropriate).
a search path setup, some of these hosts resolve to the wrong address.
By appending a period to the hostname, the hostname should only resolve
to what we want it to resolve to. Hopefully this doesn't break different bots.
Also add more info to failure message to aid debugging test failure.
a search path setup, some of these hosts resolve to the wrong address.
By appending a period to the hostname, the hostname should only resolve
to what we want it to resolve to. Hopefully this doesn't break different bots.
clean up files and directories the tests often leave behind by
mistake. This is the first time in history I don't have a bogus
"db_home" directory after running the tests ;-)
Also worked on runtest's docstring, to say something about all the
arguments, and to document the non-obvious return values.
New functions runtest_inner() and cleanup_test_droppings() in
support of the above.
run immediately after test_file. At least 8 buildbot
boxes passed since the underlying problem got fixed,
and they all failed before the fix, so there's no point
to this anymore.
test clean up after itself appears to fix the test failures
when test_optparse follows test_file.
test_main(): Get rid of TESTFN no matter what. That's
also enough to fix the mystery failures. Doesn't hurt
to fix them twice :-)
Since test_file is implicated in mysterious test failures
when followed by test_optparse, if I had any brains I'd
look at the checkin that last changed test_file ;-)
the char buffer was requested. Now it actually returns the char buffer if
available or raises a TypeError if it isn't (as is raised for the other buffer
types if they are not present but requested).
Not a backport candidate since it does change semantics of the buffer object
(although it could be argued this is enough of a bug to bother backporting).
Give a consistent behavior for comparison and hashing of method objects
(both user- and built-in methods). Now compares the 'self' recursively.
The hash was already asking for the hash of 'self'.
from ?iga Seilnacht (sorry about the name, but Firefox
on my box can't display the first character of the name --
the SF "Unix name" is zseil).
This appears to cure the oddball intermittent leaks across
runs when running test_exceptions under -R. I'm not sure
why, but I'm too sleepy to care ;-)
The thrust of the SF patch was to remove randomness in the
pickle protocol used. I changed the patch to use
range(pickle.HIGHEST_PROTOCOL + 1), to try both pickle and
cPickle, and randomly mucked with other test lines to put
statements on their own lines.
Not a bugfix candidate (this is fiddling new-in-2.5 code).
Found them using::
find . -name '*.py' | while read i ; do grep 'def[^(]*( ' $i /dev/null ; done
find . -name '*.py' | while read i ; do grep ' ):' $i /dev/null ; done
(I was doing this all over my own code anyway, because I'd been using spaces in
all defs, so I thought I'd make a run on the Python code as well. If you need
to do such fixes in your own code, you can use xx-rename or parenregu.el within
emacs.)
string_reverse(): Simplify.
assertRaises(): Raise TestFailed on failure.
test_unpack_from(), test_pack_into(), test_pack_into_fn(): never
use `assert` to test for an expected result (it doesn't test anything
when Python is run with -O).
Move the long-winded, multiply-nested -R support out
of runtest() and into some module-level helper functions.
This makes runtest() and the -R code easier to follow.
That in turn allowed seeing some opportunities for code
simplification, and made it obvious that reglog.txt
never got closed.
46640 Patch #1454481: Make thread stack size runtime tunable.
46647 Markup fix
The first is causing many buildbots to fail test runs, and there
are multiple causes with seemingly no immediate prospects for
repairing them. See python-dev discussion.
Note that a branch can (and should) be created for resolving these
problems, like
svn copy svn+ssh://svn.python.org/python/trunk -r46640 svn+ssh://svn.python.org/python/branches/NEW_BRANCH
followed by merging rev 46647 to the new branch.
- Following Guido's comments, renamed
* pack_to -> pack_into
* recv_buf -> recv_into
* recvfrom_buf -> recvfrom_into
- Made fixes to _struct.c according to Neal Norwitz comments on the checkins
list.
- Converted some ints into the appropriate -- I hope -- ssize_t and size_t.
the output required more than one line. "Small" dicts got
displayed in seemingly random order (the hash-induced order
produced by dict.__repr__). None of this was documented.
Now pprint functions always sort dicts by key, and the docs
promise it.
This was proposed and agreed to during the PyCon 2006 core
sprint -- I just didn't have time for it before now.
but disabled then because str and unicode strings gave different
results. The implementations were repaired later during the
sprint, but the new test remained disabled.
are run in the order:
test_genexps (or any other doctest-based test)
test_struct
test_doctest
The `warnings` module needs an advertised way to save/restore
its internal filter list.
Renames functional extension module to _functools and adds a Python
functools module so that utility functions like update_wrapper can be
added easily.
Remove various dependencies on dictionary order in the standard library
tests, and one (clearly an oversight, potentially critical) in the
standard library itself - base64.py.
Remaining open issues:
* test_extcall is an output test, messy to make robust
* tarfile.py has a potential bug here, but I'm not familiar
enough with this code. Filed in as SF bug #1496501.
* urllib2.HTTPPasswordMgr() returns a random result if there is more
than one matching root path. I'm asking python-dev for
clarification...
failures on Windows buildbots, but it's hard to know how since the regrtest
failure output is useless here, and it never fails when a buildbot slave runs
test_tarfile the second time in verbose mode.
The new split functions use a preallocated list. Added tests which exceed
the preallocation size, to exercise list appends/resizes.
Also added more edge case tests.
* Added socket.recv_buf() and socket.recvfrom_buf() methods, that use the buffer
protocol (send and sendto already did).
* Added struct.pack_to(), that is the corresponding buffer compatible method to
unpack_from().
* Fixed minor typos in arraymodule.
the Need For Speed sprint coding. Includes commented out overflow tests
which will be uncommented once the code is fixed.
This test will break the 8-bit string tests because
"".replace("", "A") == "" when it should == "A"
We have a fix for it, which should be added tomorrow.
In rare cases of strings specifying true values near sys.maxint,
and oddball bases (not decimal or a power of 2), int(string, base)
could deliver insane answers. This repairs all such problems, and
also speeds string->int significantly. On my box, here are %
speedups for decimal strings of various lengths:
length speedup
------ -------
1 12.4%
2 15.7%
3 20.6%
4 28.1%
5 33.2%
6 37.5%
7 41.9%
8 46.3%
9 51.2%
10 19.5%
11 19.9%
12 23.9%
13 23.7%
14 23.3%
15 24.9%
16 25.3%
17 28.3%
18 27.9%
19 35.7%
Note that the difference between 9 and 10 is the difference between
short and long Python ints on a 32-bit box. The patch doesn't
actually do anything to speed conversion to long: the speedup is
due to detecting "unsigned long" overflow more quickly.
This is a bugfix candidate, but it's a non-trivial patch and it
would be painful to separate the "bug fix" from the "speed up" parts.
the file object in binary mode.
The Windows buildbot slaves shouldn't swap themselves to death
anymore. However, test_tarfile may still fail because of a
temp directory left behind from a previous failing run.
Windows buildbot owners may need to remove that directory
by hand.
discussion.
There are two places of documentation that still mention __context__:
Doc/lib/libstdtypes.tex -- I wasn't quite sure how to rewrite that without
spending a whole lot of time thinking about it; and whatsnew, which Andrew
usually likes to change himself.
Patch #1464708 from William McVey: fixed handling of nested comments in mail
addresses. E.g.
"Foo ((Foo Bar)) <foo@example.com>"
Fixes for both rfc822.py and email package. This patch needs to be back
ported to Python 2.3 for email 2.5.
Printing to stdout, doesn't mean the data was actually written.
It depends on the buffering, so we need to flush. This will hopefully
really fix the buildbots getting killed due to no output on the slow bots.
- Warn-raise ImportWarning when importing would have picked up a directory
as package, if only it'd had an __init__.py. This swaps two tests (for
case-ness and __init__-ness), but case-test is not really more expensive,
and it's not in a speed-critical section.
- Test for the new warning by importing a common non-package directory on
sys.path: site-packages
- In regrtest.py, silence warnings generated by the build-environment
because Modules/ (which is added to sys.path for Setup-created modules)
has 'zlib' and '_ctypes' directories without __init__.py's.
tuples. Lots to be added, still, but this will give big-memory people
something to play with in 2.5 alpha 2, and hopefully get more people to
write these tests.
uses of it in test_with.py. As a result, test_with has been skipped
(due to failing imports) on all buildbot boxes since. Alas, that's
not a test failure -- you have to pay attention to the
1 skip unexpected on PLATFORM:
test_with
kinds of output at the ends of test runs to notice that this got
broken.
It's likely that more renaming in test_with.py would be desirable.
The new char-array used in ioctl calls wasn't explicitly NUL-terminated;
quite probably the cause for the test_pty failures on Solaris that we
circumvented earlier. (I wasn't able to reproduce it with this patch, but it
has been somewhat elusive to start with.)
terminology in the alpha 1 documentation.
- "context manager" reverts to its alpha 1 definition
- the term "context specifier" goes away entirely
- contextlib.GeneratorContextManager is renamed GeneratorContext
There are still a number of changes relative to alpha 1:
- the expression in the with statement is explicitly called the
"context expression" in the language reference
- the terms 'with statement context', 'context object' or 'with
statement context' are used in several places instead of a bare
'context'. The aim of this is to avoid ambiguity in relation to the
runtime context set up when the block is executed, and the context
objects that already exist in various application domains (such as
decimal.Context)
- contextlib.contextmanager is renamed to contextfactory
This best reflects the nature of the function resulting from the
use of that decorator
- decimal.ContextManager is renamed to WithStatementContext
Simple dropping the 'Manager' part wasn't possible due to the
fact that decimal.Context already exists and means something
different. WithStatementContext is ugly but workable.
A technically unrelated change snuck into this commit:
contextlib.closing now avoids the overhead of creating a
generator, since it's trivial to implement that particular
context manager directly.
the 2005 Summer of Code).
The revision adds a number of new mailbox classes that support adding
and removing messages; these classes also support mailbox locking and
default to using email.Message instead of rfc822.Message.
The old mailbox classes are largely left alone for backward compatibility.
The exception is the Maildir class, which was present in the old module
and now inherits from the new classes. The Maildir class's interface
is pretty simple, though, so I think it'll be compatible with existing
code.
(The change to the NEWS file also adds a missing word to a different
news item, which unfortunately required rewrapping the line.)
compatibility classes in the new mailbox.py that I'll be committing in
a few minutes.
One change has been made: the tests use len(mbox) instead of len(mbox.boxes).
The 'boxes' attribute was never documented and contains some internal state
that seems unlikely to have been useful.
Python 2.4 changed ntpath.abspath to do an import
inside the function. As a result, due to Python's
import lock, anything calling abspath on Windows
(directly, or indirectly like tempfile.TemporaryFile)
hung when it was called from a thread spawned as a
side effect of importing a module.
This is a depressingly frequent problem, and
deserves a more general fix. I'm settling for
a micro-fix here because this specific one accounts
for a report of Zope Corp's ZEO hanging on Windows,
and it was an odd way to change abspath to begin
with (ntpath needs a different implementation
depending on whether we're actually running on
Windows, and the _obvious_ way to arrange for that
is not to bury a possibly-failing import _inside_
the function).
Note that if/when other micro-fixes of this kind
get made, the new Lib/test/threaded_import_hangers.py
is a convenient place to add tests for them.
and provide a substitute if the import fails, because pyclbr sees the
class definition. Changed to ignore such cases' base classes and methods,
since they will not match.
'python.org' when deciding what server to use for the timeout tests; getting
tired of seeing the test fail on all my boxes ;P This'll still allow the
test to fail for hosts in the XS4ALL network that don't have an 'xs4all'
hostname, so maybe it should use a fallback scheme instead.
exceptions that can't be raised any further, because (for instance) they
occur in __del__ methods. The coroutine tests in test_generators was
triggering this leak. Remove the leakers' testcase, and add a simpler
testcase that explicitly tests this leak to test_generators.
test_generators now no longer leaks at all, on my machine. This fix may also
solve other leaks, but my full refleakhunting run is still busy, so who
knows?
not be tracked by GC. This fixes 254 of test_generators' refleaks on my
machine, but I'm sure something else will make them come back :>
Not adding a separate test for this kind of cycle, since the existing
fib/m235 already test them in more extensive ways than any 'minimal' test
has been able to manage.
examples no longer require any explicit closing to avoid
leaking.
That the tee-based examples still do is (I think) still a
mystery. Part of the mystery is that gc.garbage remains
empty: if it were the case that some generator in a trash
cycle said it needed finalization, suppressing collection
of that cycle, that generator _would_ show up in gc.garbage.
So this is acting more like, e.g., some tp_traverse slot
isn't visiting all the pointers it should (in which case
the skipped pointer(s) would act like an external root,
silently suppressing collection of everything reachable
from it(them)).
problems: first, PyGen_NeedsFinalizing() had an off-by-one bug that
prevented it from ever saying a generator didn't need finalizing, and
second, frame objects cleared themselves in a way that caused their
owning generator to think they were still executable, causing a double
deallocation of objects on the value stack if there was still a loop
on the block stack. This revision also removes some unnecessary
close() operations from test_generators that are now appropriately
handled by the cycle collector.
an incremental encoder that must retain part of the data between calls
to the encode() method.
Fix the incremental encoder and decoder for the IDNA encoding.
This closes SF patch #1453235.
The test case came from test_generators, not test_itertools.
Ensure there's no cyclic garbage we are counting.
This is weird because it leaks, then reaches a limit:
python.exe -i test_tee.py
>>> leak()
0
[26633 refs]
>>> leak()
0
[26658 refs]
>>> leak()
0
[26683 refs]
>>> leak()
0
[26708 refs]
>>> leak()
0
[26708 refs]
>>> leak()
0
[26708 refs]
>>> leak()
0
appear. Get rid of them by nuking doctest's default DocTestRunner
instance as part of cleanup(). Also cleanup() before running the
first test repetition (the test was run once before we get into
the -R branch).
Facundo Batista
Eric Smith
Brett Cannon
Amaury Forgeot d'Arc
Christian Heimes
Jeffrey Yasskin
Raymond Hettinger
Mark Dickinson
Lars Gustäbel
Skip Montanaro
Nick Coghlan
Hye-Shik Chang
Andrew M. Kuchling
Georg Brandl
Neal Norwitz
Guido van Rossum
Gregory P. Smith
Fred Drake
Kurt B. Kaiser
Alexandre Vassalotti
Armin Rigo
Martin v. Löwis
Walter Dörwald
Bill Janssen
Travis E. Oliphant
Thomas Wouters
Collin Winter
Alex Martelli
Peter Astrand
Kristján Valur Jónsson
Barry Warsaw
Žiga Seilnacht
Tim Peters
Vinay Sajip
Jeremy Hylton
Sjoerd Mullender
Michael W. Hudson
Thomas Heller
Jack Diederich
Gustavo Niemeyer
Neil Schemenauer
Fredrik Lundh
Ka-Ping Yee
Marc-André Lemburg
Andrew MacIntyre
Bob Ippolito
Greg Ward
Phillip J. Eby
Ronald Oussoren
Martin Blais
George Yoshida
Richard Jones
Andrew Dalke
Gerhard Häring