]> granicus.if.org Git - python/log
python
22 years agoAdd local_hostname option to SMTP.__init__. If supplied, it is used
Neil Schemenauer [Sun, 24 Mar 2002 15:30:40 +0000 (15:30 +0000)]
Add local_hostname option to SMTP.__init__.  If supplied, it is used
as the fully qualified local hostname.

22 years agoTry /var/mail before trying /usr/mail. Most new systems use /var.
Neil Schemenauer [Sun, 24 Mar 2002 01:38:38 +0000 (01:38 +0000)]
Try /var/mail before trying /usr/mail.  Most new systems use /var.
This in inside the test so it really doesn't matter much.  Closes
SF patch 497097.

22 years agoAdd more tests for built-in types. Add tests for buffer objects.
Neil Schemenauer [Sun, 24 Mar 2002 01:24:54 +0000 (01:24 +0000)]
Add more tests for built-in types.  Add tests for buffer objects.
Closes SF patch 494871.

22 years agoAdd get_history_item, get_current_history_length, and redisplay functions.
Neil Schemenauer [Sun, 24 Mar 2002 01:09:04 +0000 (01:09 +0000)]
Add get_history_item, get_current_history_length, and redisplay functions.
Clarify the docstring for get_history_length.  Closes SF patch 494066.

22 years ago_PyMalloc_DebugRealloc(): simplify decl of "fresh".
Tim Peters [Sun, 24 Mar 2002 00:34:21 +0000 (00:34 +0000)]
_PyMalloc_DebugRealloc():  simplify decl of "fresh".

Assorted:  bump the serial number via a trivial new bumpserialno()
function.  The point is to give a single place to set a breakpoint when
waiting for a specific serial number.

22 years agoUse linecache for loading source code. Closes SF patch 490374.
Neil Schemenauer [Sat, 23 Mar 2002 23:51:04 +0000 (23:51 +0000)]
Use linecache for loading source code.  Closes SF patch 490374.

22 years agoAdd namespace selection for rlcompleter. Closes SF patch 490026.
Neil Schemenauer [Sat, 23 Mar 2002 23:44:51 +0000 (23:44 +0000)]
Add namespace selection for rlcompleter.  Closes SF patch 490026.

22 years agoMinor code cleanup -- no semantic changes.
Tim Peters [Sat, 23 Mar 2002 22:28:18 +0000 (22:28 +0000)]
Minor code cleanup -- no semantic changes.

22 years agoFix example for PyErr_SetFromErrno() (need to pass exception type).
Neil Schemenauer [Sat, 23 Mar 2002 20:57:11 +0000 (20:57 +0000)]
Fix example for PyErr_SetFromErrno() (need to pass exception type).

22 years agoFix wording of sys.exit docstring. Close SF bug 534113.
Neil Schemenauer [Sat, 23 Mar 2002 20:46:35 +0000 (20:46 +0000)]
Fix wording of sys.exit docstring.  Close SF bug 534113.

22 years agoFlush stdout before reading next command. Closes SF bug 526357.
Neil Schemenauer [Sat, 23 Mar 2002 20:43:59 +0000 (20:43 +0000)]
Flush stdout before reading next command.  Closes SF bug 526357.

22 years agoGrow the string buffer at a mildly exponential rate for the getc version
Neil Schemenauer [Sat, 23 Mar 2002 19:41:34 +0000 (19:41 +0000)]
Grow the string buffer at a mildly exponential rate for the getc version
of get_line.  This makes test_bufio finish in 1.7 seconds instead of 57
seconds on my machine (with Py_DEBUG defined).

Also, rename the local variables n1 and n2 to used_v_size and
total_v_size.

22 years agoGive Python a debug-mode pymalloc, much as sketched on Python-Dev.
Tim Peters [Sat, 23 Mar 2002 10:03:50 +0000 (10:03 +0000)]
Give Python a debug-mode pymalloc, much as sketched on Python-Dev.

When WITH_PYMALLOC is defined, define PYMALLOC_DEBUG to enable the debug
allocator.  This can be done independent of build type (release or debug).
A debug build automatically defines PYMALLOC_DEBUG when pymalloc is
enabled.  It's a detected error to define PYMALLOC_DEBUG when pymalloc
isn't enabled.

Two debugging entry points defined only under PYMALLOC_DEBUG:

+ _PyMalloc_DebugCheckAddress(const void *p) can be used (e.g., from gdb)
  to sanity-check a memory block obtained from pymalloc.  It sprays
  info to stderr (see next) and dies via Py_FatalError if the block is
  detectably damaged.

+ _PyMalloc_DebugDumpAddress(const void *p) can be used to spray info
  about a debug memory block to stderr.

A tiny start at implementing "API family" checks isn't good for
anything yet.

_PyMalloc_DebugRealloc() has been optimized to do little when the new
size is <= old size.  However, if the new size is larger, it really
can't call the underlying realloc() routine without either violating its
contract, or knowing something non-trivial about how the underlying
realloc() works.  A memcpy is always done in this case.

This was a disaster for (and only) one of the std tests:  test_bufio
creates single text file lines up to a million characters long.  On
Windows, fileobject.c's get_line() uses the horridly funky
getline_via_fgets(), which keeps growing and growing a string object
hoping to find a newline.  It grew the string object 1000 bytes each
time, so for a million-character string it took approximately forever
(I gave up after a few minutes).

So, also:

fileobject.c, getline_via_fgets():  When a single line is outrageously
long, grow the string object at a mildly exponential rate, instead of
just 1000 bytes at a time.

That's enough so that a debug-build test_bufio finishes in about 5 seconds
on my Win98SE box.  I'm curious to try this on Win2K, because it has very
different memory behavior than Win9X, and test_bufio always took a factor
of 10 longer to complete on Win2K.  It *could* be that the endless
reallocs were simply killing it on Win2K even in the release build.

22 years agoOnly AttributeError can be raised in this situation - on systems without
Skip Montanaro [Sat, 23 Mar 2002 05:58:52 +0000 (05:58 +0000)]
Only AttributeError can be raised in this situation - on systems without
getuid or getpid.  posix_getuid & posix_getpid never raise exceptions when
called with no args.

22 years agoimport statements only raise ImportError, right?
Skip Montanaro [Sat, 23 Mar 2002 05:55:18 +0000 (05:55 +0000)]
import statements only raise ImportError, right?

22 years agotighten up except - only ValueError can be raised in this situation
Skip Montanaro [Sat, 23 Mar 2002 05:50:17 +0000 (05:50 +0000)]
tighten up except - only ValueError can be raised in this situation

22 years agotighten up except - int() only raises ValueError
Skip Montanaro [Sat, 23 Mar 2002 05:47:31 +0000 (05:47 +0000)]
tighten up except - int() only raises ValueError

22 years agoadded RFC 2396 tests from Aaron Swartz included in bug # 450225.
Skip Montanaro [Sat, 23 Mar 2002 05:32:10 +0000 (05:32 +0000)]
added RFC 2396 tests from Aaron Swartz included in bug # 450225.
converted to use unittest

22 years agono longer needed - converted test_urlparse.py to use unittest
Skip Montanaro [Sat, 23 Mar 2002 05:29:59 +0000 (05:29 +0000)]
no longer needed - converted test_urlparse.py to use unittest

22 years agoSF bug 533234: tm_isdst > 1 Passed to strftime.
Tim Peters [Sat, 23 Mar 2002 03:26:53 +0000 (03:26 +0000)]
SF bug 533234:  tm_isdst > 1 Passed to strftime.
One more time on this turkey, but duller instead of cleverer.

Curious:  The docs say __getslice__ has been deprecated since 2.0, but
list.__getitem__ still doesn't work if you pass it a slice.  This makes
it a lot clearer to emulate a list by *being* a list <wink>.

Bugfix candidate.  Michael, just pile this patch on top of the others
that went by -- no need to try to pick these apart.

22 years agoCheck in (hopefully) corrected version of last change.
Neil Schemenauer [Sat, 23 Mar 2002 02:06:50 +0000 (02:06 +0000)]
Check in (hopefully) corrected version of last change.

22 years agoJust whitespace fiddling.
Tim Peters [Sat, 23 Mar 2002 00:28:57 +0000 (00:28 +0000)]
Just whitespace fiddling.

22 years agoBuild obmalloc.c directly instead of #include'ing from object.c.
Tim Peters [Sat, 23 Mar 2002 00:20:15 +0000 (00:20 +0000)]
Build obmalloc.c directly instead of #include'ing from object.c.
Also move all _PyMalloc_XXX entry points into obmalloc.c.

The Windows build works fine.
The Unix build is changed here (Makefile.pre.in), but not tested.
No other platform's build process has been fiddled.

22 years agoDisable the parser hacks that enabled the "yield" keyword using a future
Neil Schemenauer [Fri, 22 Mar 2002 23:53:36 +0000 (23:53 +0000)]
Disable the parser hacks that enabled the "yield" keyword using a future
statement.

22 years agoUndo last commit. It's causing the tests to file.
Neil Schemenauer [Fri, 22 Mar 2002 23:50:30 +0000 (23:50 +0000)]
Undo last commit.  It's causing the tests to file.

22 years agoDisable the parser hacks that allowed the "yield" keyword to be enabled
Neil Schemenauer [Fri, 22 Mar 2002 23:38:11 +0000 (23:38 +0000)]
Disable the parser hacks that allowed the "yield" keyword to be enabled
by a future statement.

22 years agoRemove malloc hooks.
Neil Schemenauer [Fri, 22 Mar 2002 23:20:15 +0000 (23:20 +0000)]
Remove malloc hooks.

22 years agoMatch behavior of the pickle.py module more closely.
Neil Schemenauer [Fri, 22 Mar 2002 23:02:53 +0000 (23:02 +0000)]
Match behavior of the pickle.py module more closely.

22 years agoAdd more example exceptions that unpickling can raise.
Neil Schemenauer [Fri, 22 Mar 2002 22:16:03 +0000 (22:16 +0000)]
Add more example exceptions that unpickling can raise.

22 years agoHandle os.listdir("") case correctly on Windows. Closes bug 500705.
Neil Schemenauer [Fri, 22 Mar 2002 20:51:58 +0000 (20:51 +0000)]
Handle os.listdir("") case correctly on Windows.  Closes bug 500705.

22 years agoDisallow open()ing of directories. Closes SF bug 487277.
Neil Schemenauer [Fri, 22 Mar 2002 20:38:57 +0000 (20:38 +0000)]
Disallow open()ing of directories.  Closes SF bug 487277.

22 years agobetter solution for bug #533234 courtesy of Tim.
Skip Montanaro [Fri, 22 Mar 2002 18:35:51 +0000 (18:35 +0000)]
better solution for bug #533234 courtesy of Tim.

Michael: use this version as the bugfix candidate...

22 years agoguarantee that the dst flag of synthetic "time" tuples passed to strftime
Skip Montanaro [Fri, 22 Mar 2002 18:07:49 +0000 (18:07 +0000)]
guarantee that the dst flag of synthetic "time" tuples passed to strftime
is always 0.  This closes bug #533234.

22 years agouse stat attributes instead of tuple entries
Walter Dörwald [Fri, 22 Mar 2002 17:30:38 +0000 (17:30 +0000)]
use stat attributes instead of tuple entries
and remove the unneccessary "import stat" statement.

22 years agoFix broken HTML in the head; this was reported by a user as causing Opera 6.01
Fred Drake [Fri, 22 Mar 2002 17:22:38 +0000 (17:22 +0000)]
Fix broken HTML in the head; this was reported by a user as causing Opera 6.01
to crash.  The user has reported the problem to Opera, but we still should
generate something that passes for HTML.

22 years agoAdd news about pymalloc being enabled.
Neil Schemenauer [Fri, 22 Mar 2002 17:06:59 +0000 (17:06 +0000)]
Add news about pymalloc being enabled.

22 years agoUse pymalloc for realloc() as well.
Neil Schemenauer [Fri, 22 Mar 2002 15:41:03 +0000 (15:41 +0000)]
Use pymalloc for realloc() as well.

22 years agoRevert part of previous patch: several install_* subcommands expect
Andrew M. Kuchling [Fri, 22 Mar 2002 15:35:17 +0000 (15:35 +0000)]
Revert part of previous patch: several install_* subcommands expect
    .compile to be None, and set it to true if it is.
    Caught by Pearu Peterson.

Bugfix candidate, if the previous change is accepted for
release22-maint.

22 years agoEnable pymalloc by default.
Neil Schemenauer [Fri, 22 Mar 2002 15:34:49 +0000 (15:34 +0000)]
Enable pymalloc by default.

22 years agoUse pymalloc if it's enabled.
Neil Schemenauer [Fri, 22 Mar 2002 15:33:15 +0000 (15:33 +0000)]
Use pymalloc if it's enabled.

22 years agoAdd pymalloc object memory management functions. These must be
Neil Schemenauer [Fri, 22 Mar 2002 15:28:30 +0000 (15:28 +0000)]
Add pymalloc object memory management functions.  These must be
available even if pymalloc is disabled since extension modules might use
them.

22 years agoMake PyObject_{NEW,New,Del,DEL} always use the standard malloc (PyMem_*)
Neil Schemenauer [Fri, 22 Mar 2002 15:25:18 +0000 (15:25 +0000)]
Make PyObject_{NEW,New,Del,DEL} always use the standard malloc (PyMem_*)
and not pymalloc.  Add the functions PyMalloc_New, PyMalloc_NewVar, and
PyMalloc_Del that will use pymalloc if it's enabled.   If pymalloc is
not enabled then they use the standard malloc (PyMem_*).

22 years agoGenerate with weaklink stubs, so missing routines (on MacOS 8.6 and earlier)
Jack Jansen [Fri, 22 Mar 2002 14:16:39 +0000 (14:16 +0000)]
Generate with weaklink stubs, so missing routines (on MacOS 8.6 and earlier)
don't cause import failure.

Fixes 531398, 2.2.1 candidate.

22 years ago- Weaklink InterfaceLib in _Res module
Jack Jansen [Fri, 22 Mar 2002 14:15:07 +0000 (14:15 +0000)]
- Weaklink InterfaceLib in _Res module
- forgot to pass libraryflags and stdlibraryflags on to ppc/carbon
  project generation.

First half of fix to 531398.

22 years agoEnable pymalloc by default in the Windows build.
Tim Peters [Fri, 22 Mar 2002 06:32:32 +0000 (06:32 +0000)]
Enable pymalloc by default in the Windows build.

22 years agoAdd a simple test suite for netrc.py, and remove it from test_sundry
Andrew M. Kuchling [Fri, 22 Mar 2002 02:48:57 +0000 (02:48 +0000)]
Add a simple test suite for netrc.py, and remove it from test_sundry

22 years agoSF bug 533198: Complex power underflow raises exception.
Tim Peters [Fri, 22 Mar 2002 02:48:46 +0000 (02:48 +0000)]
SF bug 533198:  Complex power underflow raises exception.
Konrad was too kind.  Not only did it raise an exception, the specific
exception it raised made no sense.  These are old bugs in complex_pow()
and friends:

1. Raising 0 to a negative power isn't a range error, it's a domain
   error, so changed c_pow() to set errno to EDOM in that case instead
   of ERANGE.

2. Changed complex_pow() to:

A. Used the Py_ADJUST_ERANGE2 macro to try to clear errno of a spurious
   ERANGE error due to underflow in the libm pow() called by c_pow().

B. Produced different exceptions depending on the errno value:
   i) For errno==EDOM, raise ZeroDivisionError instead of ValueError.
      This is for consistency with the non-complex cases 0.0**-2 and
      0**-2 and 0L**-2.
   ii) For errno==ERANGE, raise OverflowError.

Bugfix candidate.

22 years ago[Bug #532115] netrc module was broken
Andrew M. Kuchling [Fri, 22 Mar 2002 02:46:41 +0000 (02:46 +0000)]
[Bug #532115]  netrc module was broken
   * 'macdef' (macro definition) wasn't parsed correctly
   * account value not reset for a subsequent 'default' line
   * typo: 'whitepace' -> 'whitespace'

Bugfix candidate.

22 years ago[Bug #532136] Change path in README, as suggested by Ralph Corderoy
Andrew M. Kuchling [Thu, 21 Mar 2002 23:52:20 +0000 (23:52 +0000)]
[Bug #532136] Change path in README, as suggested by Ralph Corderoy

Bugfix candidate.

22 years agoAdd missing Boolean options
Andrew M. Kuchling [Thu, 21 Mar 2002 23:46:54 +0000 (23:46 +0000)]
Add missing Boolean options
Remove unused no_compile flag
Initialize the Boolean attribute .compile to 0 instead of None

Bugfix candidate.

22 years agoAdd unlisted Boolean options. Thomas H., can you please check that I
Andrew M. Kuchling [Thu, 21 Mar 2002 23:44:01 +0000 (23:44 +0000)]
Add unlisted Boolean options.  Thomas H., can you please check that I
    got this right?

Bugfix candidate, unless Thomas notes a problem.

22 years ago[Bug #517451] bdist_rpm didn't list all of its Boolean options.
Andrew M. Kuchling [Thu, 21 Mar 2002 23:27:54 +0000 (23:27 +0000)]
[Bug #517451] bdist_rpm didn't list all of its Boolean options.
   (Someone should check the other commands for this same error.)

Bugfix candidate.

22 years agoAdded an open_pathname() method which opens a resource file by pathname,
Jack Jansen [Thu, 21 Mar 2002 22:38:32 +0000 (22:38 +0000)]
Added an open_pathname() method which opens a resource file by pathname,
possibly converting from AppleSingle.

22 years agoFor reasons I don't fully understand we sometimes get unexpected events
Jack Jansen [Thu, 21 Mar 2002 22:36:57 +0000 (22:36 +0000)]
For reasons I don't fully understand we sometimes get unexpected events
in MachoPython. As we don't have MacOS.HandleEvent() we drop these on
the floor (with a print).

22 years agoSolve sync() signature conflict differently: manually return 0 if
Jack Jansen [Thu, 21 Mar 2002 21:09:36 +0000 (21:09 +0000)]
Solve sync() signature conflict differently: manually return 0 if
we have GUSI's void sync().

22 years agoNew checkin to convert from BinHex to AppleSingle encoding.
Jack Jansen [Thu, 21 Mar 2002 20:18:32 +0000 (20:18 +0000)]
New checkin to convert from BinHex to AppleSingle encoding.

22 years agoPatch #532729: check for sem_init in -lrt.
Martin v. Löwis [Thu, 21 Mar 2002 15:10:58 +0000 (15:10 +0000)]
Patch #532729: check for sem_init in -lrt.

22 years agoFix grammar
Neal Norwitz [Thu, 21 Mar 2002 12:58:54 +0000 (12:58 +0000)]
Fix grammar

22 years ago[Apply SF patch #504943]
Walter Dörwald [Thu, 21 Mar 2002 10:38:40 +0000 (10:38 +0000)]
[Apply SF patch #504943]

This patch makes it possible to pass Warning instances as the first
argument to warnings.warn. In this case the category argument
will be ignored. The message text used will be str(warninginstance).

22 years agoDo not insert characters for unicode-escape decoders if the error mode
Martin v. Löwis [Thu, 21 Mar 2002 08:55:28 +0000 (08:55 +0000)]
Do not insert characters for unicode-escape decoders if the error mode
is "ignore". Fixes #529104.

22 years ago[Apply patch #500457 from the PyXML tracker]
Andrew M. Kuchling [Wed, 20 Mar 2002 23:56:34 +0000 (23:56 +0000)]
[Apply patch #500457 from the PyXML tracker]
Add iterator support to pulldom.DOMEventStream

New feature, so not a bugfix candidate (though it should be safe for inclusion)

22 years agoRemove compiler warnings on Solaris 8.
Neal Norwitz [Wed, 20 Mar 2002 21:32:07 +0000 (21:32 +0000)]
Remove compiler warnings on Solaris 8.
Can go into 2.2.x, but not necessary.

22 years agoChange raw "except:" constructs to pass on KeyboardInterrupt.
Tim Peters [Wed, 20 Mar 2002 19:32:03 +0000 (19:32 +0000)]
Change raw "except:" constructs to pass on KeyboardInterrupt.

Bugfix candidate?  Don't know -- never bothered me, but it's minor
either way.

22 years agoSF# 522426, add doc for common parameter for filecmp.cmpfiles()
Neal Norwitz [Wed, 20 Mar 2002 18:55:09 +0000 (18:55 +0000)]
SF# 522426, add doc for common parameter for filecmp.cmpfiles()

22 years agoMake GzipFile an iterator. Closes bug #532621.
Neil Schemenauer [Wed, 20 Mar 2002 18:36:00 +0000 (18:36 +0000)]
Make GzipFile an iterator.  Closes bug #532621.

22 years agoArrange to export the _PyMalloc_{Malloc, Realloc, Free} entry points. On
Tim Peters [Wed, 20 Mar 2002 04:02:31 +0000 (04:02 +0000)]
Arrange to export the _PyMalloc_{Malloc, Realloc, Free} entry points.  On
Windows some modules are considered (by me, and I don't care what anyone
else thinks about this <wink>) to be part of "the core" despite that they
happen to be compiled into separate DLLs (the "to DLL or not to DLL?"
question on Windows is nearly arbitrary).  Making the pymalloc entry
points available to them allows the Windows build to complete without
incident when WITH_PYMALLOC is #define'd.

Note that this isn't unprecedented.  Other "private API" functions we
export include _PySequence_IterSearch, _PyEval_SliceIndex, _PyCodec_Lookup,
_Py_ZeroStruct, _Py_TrueStruct, _PyLong_New and _PyModule_Clear.

22 years agoA faster version of the find_prefix_at_end() function (that I found in the
Andrew M. Kuchling [Wed, 20 Mar 2002 02:22:58 +0000 (02:22 +0000)]
A faster version of the find_prefix_at_end() function (that I found in the
    last Medusa release)

Should be safe as a bugfix candidate, though it's not fixing a bug.

22 years ago[Bug #528914] PyTraceBack_Store/Fetch were deleted in 1997, but their
Andrew M. Kuchling [Tue, 19 Mar 2002 16:02:35 +0000 (16:02 +0000)]
[Bug #528914] PyTraceBack_Store/Fetch were deleted in 1997, but their
              prototypes remain.  Noted by Yakov Markovitch.

Bugfix candidate.

22 years agoExpand LINKCC in configure.in. Suggested in bug report #529713.
Martin v. Löwis [Tue, 19 Mar 2002 15:15:32 +0000 (15:15 +0000)]
Expand LINKCC in configure.in. Suggested in bug report #529713.

22 years agoAdjust some poor wording in the text that explains what events are used
Fred Drake [Tue, 19 Mar 2002 14:37:44 +0000 (14:37 +0000)]
Adjust some poor wording in the text that explains what events are used
for (reported by Keith Briggs).
Wrap some very long lines.

22 years agoClarify that copy_reg.pickle() is not intended for use with "classic" classes.
Fred Drake [Tue, 19 Mar 2002 03:33:33 +0000 (03:33 +0000)]
Clarify that copy_reg.pickle() is not intended for use with "classic" classes.
This was stated before, but a minor grammatical error made it difficult to be
sure of the meaning.
This closes SF bug #530143.

22 years agoRecord 2.2.1c1 release date.
Tim Peters [Tue, 19 Mar 2002 02:12:10 +0000 (02:12 +0000)]
Record 2.2.1c1 release date.

22 years ago[Bug #531616] Make HTTPS work again by adding a sendall method to the
Andrew M. Kuchling [Mon, 18 Mar 2002 22:51:48 +0000 (22:51 +0000)]
[Bug #531616] Make HTTPS work again by adding a sendall method to the
FakeSocket class.  Without it, the sendall() call got the method on
the underlying socket object, and that messed up SSL.

Does httplib use other methods of sockets that FakeSocket doesn't support?
Someone should take a look...  (I'll try to give it a once-over.)

2.2.1 bugfix candidate.

22 years agoTo make 'urllib.py -t' run again, change FTP URL to a file that actually
Andrew M. Kuchling [Mon, 18 Mar 2002 22:18:46 +0000 (22:18 +0000)]
To make 'urllib.py -t' run again, change FTP URL to a file that actually
   exists.

22 years agouse -kb on idle/Icons/minusnode.gif ("cvs up" kept checking it out all
Fredrik Lundh [Mon, 18 Mar 2002 21:43:04 +0000 (21:43 +0000)]
use -kb on idle/Icons/minusnode.gif ("cvs up" kept checking it out all
the time...)

22 years agoDrop the PyCore_* memory API.
Neil Schemenauer [Mon, 18 Mar 2002 21:06:21 +0000 (21:06 +0000)]
Drop the PyCore_* memory API.

22 years agoRe-enable GC of generator objects.
Neil Schemenauer [Mon, 18 Mar 2002 20:45:09 +0000 (20:45 +0000)]
Re-enable GC of generator objects.

22 years agoRe-enable GC of method objects.
Neil Schemenauer [Mon, 18 Mar 2002 20:44:53 +0000 (20:44 +0000)]
Re-enable GC of method objects.

22 years agoRe-enable GC of iter objects.
Neil Schemenauer [Mon, 18 Mar 2002 20:43:51 +0000 (20:43 +0000)]
Re-enable GC of iter objects.

22 years ago(py-temp-directory): Add /var/tmp to the list of directories this
Barry Warsaw [Mon, 18 Mar 2002 18:53:56 +0000 (18:53 +0000)]
(py-temp-directory): Add /var/tmp to the list of directories this
searches.  This is added after /tmp.  Closes SF bug #505488, except
that /var/tmp comes after /tmp instead of the patch's suggestion of
putting it before /usr/tmp.

22 years agoRemove extraneous #define as per effbot's instructions in:
Neal Norwitz [Mon, 18 Mar 2002 18:46:14 +0000 (18:46 +0000)]
Remove extraneous #define as per effbot's instructions in:

[ 530285 ] redefining SRE_CODE in Modules/sre.h

22 years agoSimpilify PyCore_* macros by assuming the function prototypes for
Neil Schemenauer [Mon, 18 Mar 2002 18:13:41 +0000 (18:13 +0000)]
Simpilify PyCore_* macros by assuming the function prototypes for
malloc() and free() don't change.

22 years agoFix up unescaped tilde; reported by several people.
Fred Drake [Mon, 18 Mar 2002 16:47:14 +0000 (16:47 +0000)]
Fix up unescaped tilde; reported by several people.

22 years agoRemove extra verb; reported by Detlef Lannert.
Fred Drake [Mon, 18 Mar 2002 16:45:01 +0000 (16:45 +0000)]
Remove extra verb; reported by Detlef Lannert.

22 years agoLurking bug found by patch for 531291: FSSpecs should be passed to
Jack Jansen [Mon, 18 Mar 2002 15:41:32 +0000 (15:41 +0000)]
Lurking bug found by patch for 531291: FSSpecs should be passed to
Py_BuildValue by address, not by value.

2.2.1 candidate.

22 years agoSome of the newer routines return an OSErr, like most toolbox calls, in stead of...
Jack Jansen [Mon, 18 Mar 2002 15:31:08 +0000 (15:31 +0000)]
Some of the newer routines return an OSErr, like most toolbox calls, in stead of depending on ResError(). Second half of fix for #531291.

2.2.1 candidate.

22 years agoSome structures should be passed to Py_BuildValue by reference, not by value,
Jack Jansen [Mon, 18 Mar 2002 15:24:22 +0000 (15:24 +0000)]
Some structures should be passed to Py_BuildValue by reference, not by value,
notably FSSpec and FSRef objects. First half of fix for #531291.

2.2.1 candidate.

22 years agoamk's fix attached to
Michael W. Hudson [Mon, 18 Mar 2002 13:06:00 +0000 (13:06 +0000)]
amk's fix attached to

[ 516299 ] urlparse can get fragments wrong

22 years agoPatch #495598: add an -q (quiet) option to pycompile.
Martin v. Löwis [Mon, 18 Mar 2002 12:44:08 +0000 (12:44 +0000)]
Patch #495598: add an -q (quiet) option to pycompile.

22 years agoCorrectly set default entry in all cases.
Martin v. Löwis [Mon, 18 Mar 2002 10:43:18 +0000 (10:43 +0000)]
Correctly set default entry in all cases.

22 years agoPatch #499513: use readline() instead of readlines(). Removed the
Martin v. Löwis [Mon, 18 Mar 2002 10:41:20 +0000 (10:41 +0000)]
Patch #499513: use readline() instead of readlines(). Removed the
unnecessary redirection limit code which is already in FancyURLopener.

22 years agoPatch #512466: Script to move faqwiz entries.
Martin v. Löwis [Mon, 18 Mar 2002 08:56:23 +0000 (08:56 +0000)]
Patch #512466: Script to move faqwiz entries.

22 years agoPatch #525870: Avoid duplicate stat calls, use st_ attributes.
Martin v. Löwis [Mon, 18 Mar 2002 08:37:19 +0000 (08:37 +0000)]
Patch #525870: Avoid duplicate stat calls, use st_ attributes.

22 years agoFix for SF bug 528132 (Armin Rigo): classmethod().__get__() segfault
Guido van Rossum [Mon, 18 Mar 2002 03:09:06 +0000 (03:09 +0000)]
Fix for SF bug 528132 (Armin Rigo): classmethod().__get__() segfault

The proper fix is not quite what was submitted; it's really better to
take the class of the object passed rather than calling PyMethod_New
with NULL pointer args, because that can then cause other core dumps
later.

I also added a testcase for the fix to classmethods() in test_descr.py.

I've already applied this to the 2.2 branch.

22 years agodump the string exception fallback for anydbm.error.
Skip Montanaro [Mon, 18 Mar 2002 03:07:20 +0000 (03:07 +0000)]
dump the string exception fallback for anydbm.error.

22 years agoinitial test_anydbm module - based on test_dumbdbm.py
Skip Montanaro [Mon, 18 Mar 2002 03:00:37 +0000 (03:00 +0000)]
initial test_anydbm module - based on test_dumbdbm.py

22 years agofurther work on dynamic reconfiguration;
Steven M. Gava [Mon, 18 Mar 2002 02:38:44 +0000 (02:38 +0000)]
further work on dynamic reconfiguration;
keybindings

22 years agonetrc will now raise a more predictable exception when $HOME is not set
Paul Prescod [Mon, 18 Mar 2002 02:13:48 +0000 (02:13 +0000)]
netrc will now raise a more predictable exception when $HOME is not set
(as it is often not on Windows). The code was always designed so that it
would raise an IOError if there was no .netrc. But if there was no $HOME
it would return a KeyError which would be somewhat unexpected for code
that didn't know the algorithm it used to find .netrc. The particular
code that triggered this problem for me was ftpmirror.py which handled
the IOError gracefully, but not the KeyError.

22 years agoadded note that xmlrpclib won't marshal instances of subclasses of the
Skip Montanaro [Sun, 17 Mar 2002 23:15:02 +0000 (23:15 +0000)]
added note that xmlrpclib won't marshal instances of subclasses of the
builtin types

22 years agorestructure a bit to not rely on test case execution ordering
Skip Montanaro [Sun, 17 Mar 2002 23:03:42 +0000 (23:03 +0000)]
restructure a bit to not rely on test case execution ordering
add test case for bug #482460