This adds support for instance to the constructor (instances
have to define __str__ and can return Unicode objects via that
hook; string return values are decoded into Unicode using the
current default encoding).
The common technique for printing out a pointer has been to cast to a long
and use the "%lx" printf modifier. This is incorrect on Win64 where casting
to a long truncates the pointer. The "%p" formatter should be used instead.
The problem as stated by Tim:
> Unfortunately, the C committee refused to define what %p conversion "looks
> like" -- they explicitly allowed it to be implementation-defined. Older
> versions of Microsoft C even stuck a colon in the middle of the address (in
> the days of segment+offset addressing)!
The result is that the hex value of a pointer will maybe/maybe not have a 0x
prepended to it.
Notes on the patch:
There are two main classes of changes:
- in the various repr() functions that print out pointers
- debugging printf's in the various thread_*.h files (these are why the
patch is large)
Closes SourceForge patch #100505.
This patch fixes possible overflow in the use of
PyOS_GetLastModificationTime in getmtime.c and Python/import.c.
Currently PyOS_GetLastModificationTime returns a C long. This can
overflow on Win64 where sizeof(time_t) > sizeof(long). Besides it
should logically return a time_t anyway (this patch changes this).
As well, import.c uses PyOS_GetLastModificationTime for .pyc
timestamping. There has been recent discussion about the .pyc header
format on python-dev. This patch adds oveflow checking to import.c so
that an exception will be raised if the modification time
overflows. There are a few other minor 64-bit readiness changes made
to the module as well:
- size_t instead of int or long for function-local buffer and string
length variables
- one buffer overflow check was added (raises an exception on possible
overflow, this overflow chance exists on 32-bit platforms as well), no
other possible buffer overflows existed (from my analysis anyway)
Closes SourceForge patch #100509.
The common technique for printing out a pointer has been to cast to a long
and use the "%lx" printf modifier. This is incorrect on Win64 where casting
to a long truncates the pointer. The "%p" formatter should be used instead.
The problem as stated by Tim:
> Unfortunately, the C committee refused to define what %p conversion "looks
> like" -- they explicitly allowed it to be implementation-defined. Older
> versions of Microsoft C even stuck a colon in the middle of the address (in
> the days of segment+offset addressing)!
The result is that the hex value of a pointer will maybe/maybe not have a 0x
prepended to it.
Notes on the patch:
There are two main classes of changes:
- in the various repr() functions that print out pointers
- debugging printf's in the various thread_*.h files (these are why the
patch is large)
Closes SourceForge patch #100505.
This patch fixes a problem on AIX with the signed int case code in
getargs.c, after Trent Mick's intervention about MIN/MAX overflow
checks. The AIX compiler/optimizer generates bogus code with the
default flags "-g -O" causing test_builtin to fail: int("10", 16) <>
16L. Swapping the two checks in the signed int code makes the problem
go away.
Also, make the error messages fit in 80 char lines in the
source.
The depth field was never decremented inside w_object(), and it was
never initialized in PyMarshal_WriteObjectToFile().
This caused imports from .pyc files to fil mysteriously when the .pyc
file was written by the broken code -- w_object() would bail out
early, but PyMarshal_WriteObjectToFile() doesn't check the error or
return an error code, and apparently the marshalling code doesn't call
PyErr_Check() either. (That's a separate patch if I feel like it.)
Various small fixes to the builtin module to ensure no buffer
overflows.
- chunk #1:
Proper casting to ensure no truncation, and hence no surprises, in the
comparison.
- chunk #2:
The id() function guarantees a unique return value for different
objects. It does this by returning the pointer to the object. By
returning a PyInt, on Win64 (sizeof(long) < sizeof(void*)) the pointer
is truncated and the guarantee may be proven false. The appropriate
return function is PyLong_FromVoidPtr, this returns a PyLong if that
is necessary to return the pointer without truncation.
[GvR: note that this means that id() can now return a long on Win32
platforms. This *might* break some code...]
- chunk #3:
Ensure no overflow in raw_input(). Granted the user would have to pass
in >2GB of data but it *is* a possible buffer overflow condition.
As I really do not have anything better to do at the moment, I have written
a patch to Python/marshal.c that prevents Python dumping core when trying
to marshal stack bustingly deep (or recursive) data structure.
It just throws an exception; even slightly clever handling of recursive
data is what pickle is for...
[Fred Drake:] Moved magic constant 5000 to a #define.
This closes SourceForge patch #100645.
the number of children of a node exceeds the max possible value for
the short that is used to count them. The Python runtime converts
this parser error into the SyntaxError "expression too long."
module and into _exceptions.c. This includes all the PyExc_* globals,
the bltin_exc table, init_class_exc(), fini_instances(),
finierrors().
Renamed _PyBuiltin_Init_1() to _PyBuiltin_Init() since the two phase
initializations are necessary any more.
Removed as obsolete _PyBuiltin_Init_2(), _PyBuiltin_Fini_1() and
_PyBuiltin_Fini_2().
need two phase init or fini of the builtin module. Change the call of
_PyBuiltin_Init_1() to _PyBuiltin_Init(). Add a call to
init_exceptions().
Py_Finalize(): Don't call _PyBuiltin_Fini_1(). Instead call
fini_exceptions() but move this to before the thread state is
cleared.
Limit the 'b' formatter of PyArg_ParseTuple to valid values of an unsigned
char, i.e. [0,UCHAR_MAX]. It is expected that this is the common usage of 'b'.
An OverflowError is raised if the parsed value is outside this range.
Changes the 'b', 'h', and 'i' formatters in PyArg_ParseTuple to raise an
Overflow exception if they overflow (previously they just silently
overflowed).
Changes by Guido: always accept values [0..255] (in addition to
[CHAR_MIN..CHAR_MAX]) for 'b' format; changed some spaces into tabs in
other code.
who wrote:
Here's the new version of thread_nt.h. More particular, there is a
new version of thread lock that uses kernel object (e.g. semaphore)
only in case of contention; in other case it simply uses interlocked
functions, which are faster by the order of magnitude. It doesn't
make much difference without threads present, but as soon as thread
machinery initialised and (mostly) the interpreter global lock is on,
difference becomes tremendous. I've included a small script, which
initialises threads and launches pystone. With original thread_nt.h,
Pystone results with initialised threads are twofold worse then w/o
threads. With the new version, only 10% worse. I have used this
patch for about 6 months (with threaded and non-threaded
applications). It works remarkably well (though I'd desperately
prefer Python was free-threaded; I hope, it will soon).
For more comments, read the patches@python.org archives.
For documentation read the comments in mymalloc.h and objimpl.h.
(This is not exactly what Vladimir posted to the patches list; I've
made a few changes, and Vladimir sent me a fix in private email for a
problem that only occurs in debug mode. I'm also holding back on his
change to main.c, which seems unnecessary to me.)
- When 'import exceptions' fails, don't suggest to use -v to print the traceback;
this doesn't actually work.
- Remove comment about fallback to string exceptions.
- Remove a PyErr_Occurred() check after all is said and done that can
never trigger.
- Remove static function newstdexception() which is no longer called.
Added 'u' and 'u#' tags for PyArg_ParseTuple - these turn a
PyUnicodeObject argument into a Py_UNICODE * buffer, or a Py_UNICODE *
buffer plus a length with the '#'. Also added an analog to 'U'
for Py_BuildValue.
return 0 (exceptions don't match). This means that if an ImportError
is raised because exceptions.py can't be imported, the interpreter
will exit "cleanly" with an error message instead of just core
dumping.
PyErr_SetFromErrnoWithFilename(), PyErr_SetFromWindowsErrWithFilename():
Don't test on Py_UseClassExceptionsFlag.
are no longer supported (i.e. -X option is removed).
_PyBuiltin_Init_1(): Don't call initerrors(). This does mean that it
is possible to raise an ImportError before that exception has been
initialized, say because exceptions.py can't be found, or contains
bogosity. See changes to errors.c for how this is handled.
_PyBuiltin_Init_2(): Don't test Py_UseClassExceptionsFlag, just go
ahead and initialize the class-based standard exceptions. If this
fails, we throw a Py_FatalError.
Changed all references to the MAGIC constant to use a global
pyc_magic instead. This global is initially set to MAGIC, but can be
changed by the _PyImport_Init() function to provide for
special features implemented in the compiler which are settable
using command line switches and affect the way PYC files are
generated.
Currently this change is only done for the -U flag.
Support for the new -U command line option option:
with the option enabled the Python compiler
interprets all "..." strings as u"..." (same with r"..." and
ur"...").
Follow a suggestion in an /*XXX*/ comment [in com_add()] to speed up
compilation by using supplemental dictionaries to keep track of names
and constants, eliminating quadratic behavior. With this patch in
place, the time to import a 5000-line file with lots of constants [at
the global level] is reduced from 20 seconds to under 3 on my system.
Here's a patch which changes modsupport to add 'u' and 'u#',
to support building Unicode objects from a null-terminated
Py_UNICODE *, and a Py_UNICODE * with length, respectively.
[Conversion from 'U' to 'u' by Fred, based on python-dev comments.]
Note that the use of None for NULL values of the Py_UNICODE* value is
still in; I'm not sure of the conclusion on that issue.
remaining object references if the environment variable PYTHONDUMPREFS
exists. The default behaviour caused problems for background or
otherwise invisible processes that use the debug build of Python.
Fixed a memory leak found by Fredrik Lundh. Instead of
PyUnicode_AsUTF8String() we now use _PyUnicode_AsUTF8String() which
returns the string object without incremented refcount (and assures
that the so obtained object remains alive until the Unicode object is
garbage collected).
"""
Running "test_extcall" repeatedly results in memory leaks.
One of these can't be fixed (at least not easily!), it happens since
this code:
def saboteur(**kw):
kw['x'] = locals()
d = {}
saboteur(a=1, **d)
creates a circular reference - d['x']['d']==d
The others are due to some missing decrefs in ceval.c, fixed by the
patch attached below.
Note: I originally wrote this without the "goto", just adding the
missing decref's where needed. But I think the goto is justified in
keeping the executable code size of ceval as small as possible.
"""
[I think the circular reference is more like kw['x']['kw'] == kw. --GvR]
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.
comparing code objects. This give sless surprising results in
-Optimized code. It also sorts code objects by name, now.
[I changed the patch to hash() slightly to touch fewer lines.]
his copy of test_contains.py seems to be broken -- the lines he
deleted were already absent). Checkin messages:
New Unicode support for int(), float(), complex() and long().
- new APIs PyInt_FromUnicode() and PyLong_FromUnicode()
- added support for Unicode to PyFloat_FromString()
- new encoding API PyUnicode_EncodeDecimal() which converts
Unicode to a decimal char* string (used in the above new
APIs)
- shortcuts for calls like int(<int object>) and float(<float obj>)
- tests for all of the above
Unicode compares and contains checks:
- comparing Unicode and non-string types now works; TypeErrors
are masked, all other errors such as ValueError during
Unicode coercion are passed through (note that PyUnicode_Compare
does not implement the masking -- PyObject_Compare does this)
- contains now works for non-string types too; TypeErrors are
masked and 0 returned; all other errors are passed through
Better testing support for the standard codecs.
Misc minor enhancements, such as an alias dbcs for the mbcs codec.
Changes:
- PyLong_FromString() now applies the same error checks as
does PyInt_FromString(): trailing garbage is reported
as error and not longer silently ignored. The only characters
which may be trailing the digits are 'L' and 'l' -- these
are still silently ignored.
- string.ato?() now directly interface to int(), long() and
float(). The error strings are now a little different, but
the type still remains the same. These functions are now
ready to get declared obsolete ;-)
- PyNumber_Int() now also does a check for embedded NULL chars
in the input string; PyNumber_Long() already did this (and
still does)
Followed by:
Looks like I've gone a step too far there... (and test_contains.py
seem to have a bug too).
I've changed back to reporting all errors in PyUnicode_Contains()
and added a few more test cases to test_contains.py (plus corrected
the join() NameError).
If a non-tuple sequence is passed as the *arg, convert it to a tuple
before checking its length.
If named keyword arguments are used in combination with **kwargs, make
a copy of kwargs before inserting the new keys.
the return value of PySequence_Length(). If an exception occurred,
the returned length will be -1. Make sure this doesn't get obscurred,
and that the bogus length isn't used.
executive summary:
Instead of typing 'apply(f, args, kwargs)' you can type 'f(*arg, **kwargs)'.
Some file-by-file details follow.
Grammar/Grammar:
simplify varargslist, replacing '*' '*' with '**'
add * & ** options to arglist
Include/opcode.h & Lib/dis.py:
define three new opcodes
CALL_FUNCTION_VAR
CALL_FUNCTION_KW
CALL_FUNCTION_VAR_KW
Python/ceval.c:
extend TypeError "keyword parameter redefined" message to include
the name of the offending keyword
reindent CALL_FUNCTION using four spaces
add handling of sequences and dictionaries using extend calls
fix function import_from to use PyErr_Format
The attached patch set includes a workaround to get Python with
Unicode compile on BSDI 4.x (courtesy Thomas Wouters; the cause
is a bug in the BSDI wchar.h header file) and Python interfaces
for the MBCS codec donated by Mark Hammond.
Also included are some minor corrections w/r to the docs of
the new "es" and "es#" parser markers (use PyMem_Free() instead
of free(); thanks to Mark Hammond for finding these).
The unicodedata tests are now in a separate file
(test_unicodedata.py) to avoid problems if the module cannot
be found.
Attached you find the latest update of the Unicode implementation.
The patch is against the current CVS version.
It includes the fix I posted yesterday for the core dump problem
in codecs.c (was introduced by my previous patch set -- sorry),
adds more tests for the codecs and two new parser markers
"es" and "es#".
Andy Robinson noted a core dump in the codecs.c file. This
was introduced by my latest patch which fixed a memory leak
in codecs.c. The bug causes all successful codec lookups to fail.
Attached you find an update of the Unicode implementation.
The patch is against the current CVS version. I would appreciate
if someone with CVS checkin permissions could check the changes
in.
The patch contains all bugs and patches sent this week and also
fixes a leak in the codecs code and a bug in the free list code
for Unicode objects (which only shows up when compiling Python
with Py_DEBUG; thanks to MarkH for spotting this one).
Added wrapping macros to dictobject.c, listobject.c, tupleobject.c,
frameobject.c, traceback.c that safely prevends core dumps
on stack overflow. Macros and functions in object.c, object.h.
The method is an "elevator destructor" that turns cascading
deletes into tail recursive behavior when some limit is hit.
* Changes to a recent patch by Chris Tismer to errors.c. Chris' patch
always used FormatMessage() to get the error message passing the error code
from errno - but errno and FormatMessage use a different numbering scheme.
The main reason the patch looked OK was that ENOFILE==ERROR_FILE_NOT_FOUND -
but that is about the only shared error code :-). The MS CRT docs tell you
to use _sys_errlist()/_sys_nerr. My patch does also this, and adds a very
similar function specifically for win32 error codes.
PR#175 -- when exec is passed a code object, it didn't sync the locals
from the dictionary back into their fast representation.
Also took the time to remove some repetitive code there and to do the
syncing even when an exception is raised (since a partial effect
should still be synced).
* in import.c, #ifdef out references to dynamic loading based on
HAVE_DYNAMIC_LOADING
* clean out the platform-specific crud from importdl.c.
[ maybe fold this function into import.c and drop the importdl.c file? Greg.]
* change GetDynLoadFunc's "funcname" parameter to "shortname". change
"name" to "fqname" for clarification.
* each GetDynLoadFunc now creates its own funcname value.
WARNING: as I mentioned previously, we may run into an issue with a
missing "_" on some platforms. Testing will show this pretty quickly,
however.
* move pathname munging into dynload_shlib.c
Here's a patch that avoids a warning caused by the "const char* pathname"
declaration for _PyImport_GetDynLoadFunc (in dynload_aix). The "aix_load"
function's 1st arg is prototyped as "char *pathname".
file per platform (really: per style of Dl API; e.g. all platforms
using dlopen() are grouped together in dynload_shlib.c.).
This is part of a set of patches by Greg Stein.
Duzan, for AIX, to support C++ objects with static initializers, when
using the genuine IBM C++ compiler (namely xlC/xlC_r).
See accompanying patches to configure.in and acconfig.h.
not as descriptive as what Barry suggests, but this also catches the
(in my opinion important) case where some other C code besides apply()
constructs a kwdict that doesn't have the right format. All the other
possibilities of getting it wrong (non-dict, wrong keywords etc) are
already caught so this makes sense to check here.
For a long time I've seen absurd tracebacks under -O (e.g., negative
line numbers), but very rarely. Since I was looking at tracebacks
anyway, thought I'd track it down. Turns out to be Guido's only
predictable blind spot <wink -- "char" is signed on some non-GvR
systems>. Patch follows.
tracefunc (or profilefunc -- we're not sure which), zap the global
trace and profile funcs so that we can't get into recursive loop when
instantiating the resulting class based exception.
"""
Following up Robin Dunn's troubles with freeze, here's a patch that
fixes an oddity regarding the import logic of shared modules on AIX.
Symbol resolution of shared modules is now handled properly for the cases
when the python library is linked to a binary with an arbitrary name.
This includes the standard python[version] executable, but also applications
that are embedding the python core (i.e. linked with libpython[version].a,
the latter being static or shared).
"""
Introduce a new builtin exception, UnboundLocalError, raised when ceval.c
tries to retrieve or delete a local name that isn't bound to a value.
Currently raises NameError, which makes this behavior a FAQ since the same
error is raised for "missing" global names too: when the user has a global
of the same name as the unbound local, NameError makes no sense to them.
Even in the absence of shadowing, knowing whether a bogus name is local or
global is a real aid to quick understanding.
Example:
D:\src\PCbuild>type local.py
x = 42
def f():
print x
x = 13
return x
f()
D:\src\PCbuild>python local.py
Traceback (innermost last):
File "local.py", line 8, in ?
f()
File "local.py", line 4, in f
print x
UnboundLocalError: x
D:\src\PCbuild>
Note that UnboundLocalError is a subclass of NameError, for compatibility
with existing class-exception code that may be trying to catch this as a
NameError. Unfortunately, I see no way to make this wholly compatible
with -X (see comments in bltinmodule.c): under -X, [UnboundLocalError
is an alias for NameError --GvR].
[The ceval.c patch differs slightly from the second version that Tim
submitted; I decided not to raise UnboundLocalError for DELETE_NAME,
only for DELETE_LOCAL. DELETE_NAME is only generated at the module
level, and since at that level a NameError is raised for referencing
an undefined name, it should also be raised for deleting one.]
We occasionally received reports from people getting "invalid tstate"
crashes (this is a fatal error in PyThreadState_Delete()). Finally
several people were able to reproduce it reliably and Tim Peters
discovered that there is a race condition when multiple threads are
calling this function without holding the global interpreter lock (the
function may be called without holding that).
Solved the race condition by adding a lock around the mutating uses of
interp->tstate_head. Tim and Jonathan Giddy have run tests that make
it likely that this fixes the crashes -- although Tim hasn't heard
from the person who reported the original problem.
ExtensionClasses in isinstance() and issubclass().
- abstract instance and class protocols are used *only* in those
cases that would generate errors before the patch. That is, there's
no penalty for the normal case.
- instance protocol: an object smells like an instance if it
has a __class__ attribute that smells like a class.
- class protocol: an object smells like a class if it has a
__bases__ attribute that is a tuple with elements that
smell like classes (although not all elements may actually get
sniffed ;).
man pages suggest that the proper thing to do is to add THR_NEW_LWP to
the flags on thr_create(), and that there really isn't a downside, so
I'll do that.
"""
Spec says that on success pthread_create returns 0. It does not say
that an error code will be < 0. Linux glibc2 pthread_create() returns
ENOMEM (12) when one exceed process limits. (It looks like it should
return EAGAIN, but that's another story.)
For reference, see:
http://www.opengroup.org/onlinepubs/7908799/xsh/pthread_create.html
"""
[I have a feeling that similar bugs were fixed before; perhaps someone
could check that all error checks no check for != 0?]
xrange(), especially for platforms where int and long are different
sizes (so sys.maxint isn't actually the theoretical limit for the
length of a list, but the largest C int is -- sys.maxint is the
largest Python int, which is actually a C long).
Tim Peters
Marc-André Lemburg
Jack Jansen
Barry Warsaw
Guido van Rossum
Fred Drake
Jeremy Hylton
Fredrik Lundh
Mark Hammond
Andrew M. Kuchling