New submission from James G. sack (jim):
The streamreader in utf_8_sig.py fails when asked to read a specified
bytelength of data that ends up in the middle of a multibyte utf8 code.
I will attached a atandalone unittest (which does work from autotest, but
doesn't use test_support),
Changes by James G. sack (jim):
Added file: http://bugs.python.org/file8750/test_utf8sig_stream.py
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1444
__
___
Python-bugs-list
bwpy added the comment:
I can still reproduce this on FreeBSD 7.0 BETA2.
Python version is
Python 2.5.1 (r251:54863, Nov 14 2007, 13:09:04)
[GCC 4.2.1 20070719 [FreeBSD]] on freebsd7
--
nosy: +bwpy
versions: +Python 2.5
_
Tracker [EMAIL PROTECTED]
James G. sack (jim) added the comment:
re: msg57041, I'm sorry if I gave the wrong impression about interacting
with other programs. I started this feature request with some half-baked
thinking, which I tried to revise in my second post.
Anyway I'm most interested right now in lobbying for a
James G. sack (jim) added the comment:
Oops, it looks like my patch may have broken test_partial in test_codecs. I
will try to figure out what the test_partial does in the next day or so,
unless someone else can add some insignt in the meantime.
.jim
__
James G. sack (jim) added the comment:
One additional clue: test_codecs succeeds in verbose mode but fails in non-
verbose mode (autotest verbosity) .. I think. My eyes are getting
blurry. More tomorrow, I guess.
..j
__
Tracker [EMAIL PROTECTED]
New submission from Duncan Booth:
The following code throws a SystemError exception. cell_get_contents in
Objects\cellobject.c should check for a null op-ob_ref value and throw
an appropriate exception.
def oops():
def f(): cell
f.func_closure[0].cell_contents
cell =
David Binger added the comment:
The one line patch below makes
import parser; parser.sequence2st(parser.suite(class A(object):
pass).tolist())
work. It puts the parsermodule's validation back in sync
with the Python3 grammar for this rule of the grammar.
This bug is a serious problem for me.
Walter Dörwald added the comment:
jgsack wrote:
If codec utf_8 or utf_8_sig were to accept input with or without the
3-byte BOM, and write it as currently specified without/with the BOM
respectively, then _I_ can reread again with either utf_8 or utf_8_sig.
That's exactly what the utf_8_sig
Walter Dörwald added the comment:
For utf16, (arguably) a missing BOM should merely assume machian
endianess.
For utf_16_le, utf_16_be input, both should accept discard a BOM.
On output, I'm not sure; maybe all should write a BOM unless passed a flag
signifying no bom?
Or to preserve
New submission from zouguangxian:
It seems that the directory information of MSVC8 *just* can be got from
environment variable instead of registry.
This patch make me compile pywin32 with MSVC8(VS 2005).
--
files: msvccompiler.py.diff
messages: 57530
nosy: weck
severity: normal
New submission from zouguangxian:
with vs 2003, msi.py get msvcr71.dll from msm. but with vs 2005, It's
better to extract msvcr80.dll from %VCINTALLDIR%\redist\x86
\Microsoft.VC80.CRT\. In addition, it seems to extract file from
Microsoft_VC80_CRT_x86.msm need upgrade MSI to 3.1, i am not
New submission from zouguangxian:
modulator may be outdated. i made a changement to make it use the new
feature of PyTypeObject in Python2.5.
for example, to support members, methods, new, init and etc.
--
components: Demos and Tools
files: modulator.patch
messages: 57533
nosy: weck
Christian Heimes added the comment:
I'm resetting the target version of all your MSVC patches to 2.6 + 3.0.
While we can't alter the compiler version of Python 2.5 I really like to
update to a new compiler for 2.6 and 3.0.
--
components: +Demos and Tools, Installation
keywords: +patch
Changes by Christian Heimes:
--
components: +Installation
keywords: +patch
priority: - normal
versions: +Python 2.6, Python 3.0 -Python 2.5
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1448
__
Changes by Christian Heimes:
--
components: +Installation
keywords: +patch
versions: +Python 2.6, Python 3.0 -Python 2.5
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1447
__
Changes by Christian Heimes:
--
keywords: +patch
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1444
__
___
Python-bugs-list mailing list
Unsubscribe:
Christian Heimes added the comment:
I've checked your patch. It's not enough to move to VS 2005 as the
default compiler. People expect python setup.py to work w/o opening a VS
2005 Command Prompt.
--
nosy: +tiran
__
Tracker [EMAIL PROTECTED]
Changes by Georg Brandl:
--
assignee: - doerwalter
nosy: +doerwalter
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1444
__
___
Python-bugs-list mailing list
Changes by Christian Heimes:
--
resolution: - invalid
status: open - closed
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1446
__
___
Python-bugs-list mailing list
Christian Heimes added the comment:
Thanks for your work! I'm changing the version number to Python 2.6.
It's too late to get it into Python 2.5.
Christian
PS: Next time could you please create the patch in the root of the
Python source tree? It makes it easier to apply the patch.
--
Christian Heimes added the comment:
The patch looks fine but I've a small request. Could you please use a
global var PCBUILD instead of PCBuild / PCBuild8 and change the
dependencies to use either MSVCR 71 or 80 depending on the content of
the var. This way the msi module can handle both
Christian Heimes added the comment:
I'm a bit concerned with the upgrade from DB 4.4. to 4.6 and SQLite from
3.3 to 3.5. Are the APIs backward compatible or can the Python bindings
handle the new APIs? I believe that Python 2.6 is compatible with db4.6
but I'm not sure about SQLite. Have you run
Guido van Rossum added the comment:
Can you submit a unittest that catches this? Then I can check in the fix.
--
assignee: fdrake - gvanrossum
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1144
__
Bill Janssen added the comment:
I've tried several different times to fix that warning. It appears on
some releases of gcc, and not on others. There seems to be no cast or
declaration that fixes it everywhere. More power to you if you can
find one!
Bill
On 11/15/07, Christian Heimes [EMAIL
Changes by Christian Heimes:
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1378
__
___
Python-bugs-list mailing list
Unsubscribe:
Christian Heimes added the comment:
Oh, I forgot
Added file: http://bugs.python.org/file8757/py3k_another_socket.patch
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1378
__Index: Include/longobject.h
Guido van Rossum added the comment:
Committed revision 58987.
--
status: open - closed
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1144
__
___
Python-bugs-list
Martin v. Löwis added the comment:
I'm rejecting this patch, for several reasons:
- it addresses too many issues in a single patch. Separate bug reports
need to be submitted for independent issues.
- for each issue, it fails to explain what the problem is. For example,
some libraries are
Bill Janssen added the comment:
Actually, it's some combination of the version of OpenSSL plus the
version of gcc.
Bill
On 11/15/07, Bill Janssen [EMAIL PROTECTED] wrote:
I've tried several different times to fix that warning. It appears on
some releases of gcc, and not on others. There
Christian Heimes added the comment:
Ubuntu Linux 7.10, x86, gcc version 4.1.3 20070929 (prerelease) (Ubuntu
4.1.2-16ubuntu2)
/home/heimes/dev/python/py3k/Modules/_ssl.c: In function
'_get_peer_alt_names':
/home/heimes/dev/python/py3k/Modules/_ssl.c:680: warning: passing
argument 2 of
roudkerk added the comment:
From Guido's patch:
if (!DuplicateHandle(GetCurrentProcess(), (HANDLE)handle,
GetCurrentProcess(), newhandle,
0, FALSE, DUPLICATE_SAME_ACCESS))
{
Martin v. Löwis added the comment:
tiran is correct. distutils should work without having to invoke a VS
build environment. Relying on that environment would have worked way
back to VC6 and earlier, but it would reduce the ease of use of distutils.
Rejecting the patch.
--
nosy: +loewis
Guido van Rossum added the comment:
roudkerk, can you turn that suggestion into a proper patch?
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1378
__
___
Python-bugs-list
David Binger added the comment:
Okay, here is the whole thing with a unittest that exposes the problem.
Index: Lib/test/test_parser.py
===
--- Lib/test/test_parser.py (revision 58984)
+++ Lib/test/test_parser.py (working
Martin v. Löwis added the comment:
I'm opposed to this patch. Any change to the MSI build process should
only be made when/after the default compiler for Python is changed. That
needs to be discussed on python-dev first, and I hope that the new build
infrastructure will *not* use the PCbuild8
Guido van Rossum added the comment:
Forgot to upload socket3.diff.
Added file: http://bugs.python.org/file8759/socket3.diff
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1378
__Index: Include/longobject.h
Guido van Rossum added the comment:
socket3.diff builds on socket2.diff:
- Rename socket class to Socket; add socket() as a factory function.
- Implement roudkerk's suggestion of using a duplicate socket in
makefile() to get rid of the manual reference counting. Yay!
Note: this dinterferes
Guido van Rossum added the comment:
socket2.diff makes dup() a module-level function in _socket (and hence
in socket.py) and uses that to implement fromfd() in socket.py.
No changes otherwise compared to py3k_another_socket.patch.
Added file: http://bugs.python.org/file8758/socket2.diff
Guido van Rossum added the comment:
I've reviewed your patch and merged it was some pending changes of my
own. The socket tests are passing on Windows. Great work :)
You didn't upload this though.
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1378
Brett Cannon added the comment:
And under 2.6 on OS X. I doubt this is a BSD thing but more of a
recursion depth issue.
And I don't think there is anything to fix here. The recursion depth is
a per-thread thing, and this test is blowing the C stack before the
recursion depth is reached. If
Guido van Rossum added the comment:
- I'm hoping that Bill can submit his SSL changes first.
- If we make _dup() a module-level function, we can implement fromfd()
in Python. I'll do this now.
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1378
Guido van Rossum added the comment:
In release25-maint, PyErr_Print() should be replaced with
PyErr_Clear() also.
Committed revision 58991.
_
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1031213
_
Guido van Rossum added the comment:
If you haven't checked this in by tomorrow morning, I'll submit issue
1378 (socket3.diff) first, and you'll have to do a bunch of cleanup.
Or, if you like, I can submit that now and you can do the cleanup this
afternoon. (Basically, we can dup() sockets on
MHOOO added the comment:
Yeah, cool :)
Thanks =)
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1417
__
___
Python-bugs-list mailing list
Unsubscribe:
Guido van Rossum added the comment:
Patch anyone?
--
nosy: +gvanrossum
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1445
__
___
Python-bugs-list mailing list
Bill Janssen added the comment:
I'll check it in this afternoon -- I've just got to figure out how to
run the Python clean-up tool. Then you can check in the socket3.diff
patch, then I'll make sure the SSL module works with that version of
sockets.
Bill
On 11/15/07, Guido van Rossum [EMAIL
Christian Heimes added the comment:
Weck,
if you like to help and spend some time on getting VS2008 support into
Python 2.6 and 3.0 I'm willing to assist you. You can download the Beta
2 of the VS C++ 2008 Express Edition from
http://msdn2.microsoft.com/en-us/express/future/bb421473.aspx
IMHO
Christian Heimes added the comment:
I've yet another idea for a tiny improvement. Instead of doing
newfd = _socket.dup(socket_instance.fileno())
I prefer
newfd = _socket.dup(socket_instance)
It removes some confusing magic and makes error checking on Windows
slightly easier.
Martin v. Löwis added the comment:
One issue that also needs discussion is the structure of the build
directory. It could temporarily be PCbuild9, but in the long run, it
should replace PCbuild. Apart from that, the issue is whether there
should be a flat structure as it is currently in PCbuild,
Guido van Rossum added the comment:
Thanks! To be continued in issue 1378...
--
resolution: - accepted
status: open - closed
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1451
__
Bill Janssen added the comment:
OK, it's checked in. Let's see what the Windows buildbots think :-).
Bill
On Nov 15, 2007 1:35 PM, Bill Janssen [EMAIL PROTECTED] wrote:
Bill Janssen added the comment:
I'll check it in this afternoon -- I've just got to figure out how to
run the Python
Guido van Rossum added the comment:
Hold on, socket3.diff breaks four unit tests:
test_ftplib test_poplib test_smtplib test_urllib2net
newfd = _socket.dup(socket_instance)
But that doesn't allow fromfd to work. I found some real use cases for
fromfd where the fd is passed in through some
Georg Brandl added the comment:
Shouldn't the account be closed?
--
nosy: +georg.brandl
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1446
__
___
Python-bugs-list
Marc-Andre Lemburg added the comment:
I think you have a wrong understanding of round-tripping.
In Unicode it is really irrelevant if you're using a UCS2 surrogate pair
or a UCS4 representation to describe a code point. The length of the
Unicode representation may change, but the meaning
Amaury Forgeot d'Arc added the comment:
Corrected in revision 59001, with a modified patch.
--
resolution: - fixed
status: open - closed
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1134
__
Bill Janssen added the comment:
Looks like the Python SVN cert did not get included in the patch, which
is causing a failure. I'll check that in.
__
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1451
__
Guido van Rossum added the comment:
Crys, since you apparently have working sound on Windows, could you have
a look at this? There's also an (unrelated) issue with sunau.py on
Py3k, it doesn't work (but the unittests aren't strong enough to
discover that :-).
--
assignee: - tiran
roudkerk added the comment:
Currently on Windows set_error() make use of a large array which maps
socket error numbers to error messages.
This patch (against socketmodule.c from Python 2.6) removes that
array and just lets PyErr_SetExcFromWindowsErr()
generate the message by using the Win32
Guido van Rossum added the comment:
Thanks! roudkerk's patch is submitted as revision 59004.
BTW I need to go back to the drawing board for the rest of the socket
patch here. Using dup() in makefile() doesn't work for the
ssl.SSLSocket class. Maybe the explicit reference counting is the best
Ron Adam added the comment:
It looks like the disabling of \u and \U in raw strings is done. Does
tokenize.py need to be fixed, to match?
While working on this I was able to clean up the string parsing parts of
tokenize.c, and have a separate patch with just that.
And an updated patch with
Changes by Ron Adam:
Added file: http://bugs.python.org/file8763/no_raw_escapes_patch.diff
_
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1720390
_
___
Python-bugs-list
Guido van Rossum added the comment:
I don't think tokenizer.py needs to be changed -- it never interpreted
backslashes in string literals anyway (not even in regular, non-raw
literals).
The tokenizer.c cleanup is submitted as revision 59007.
I still am not warming up towards the no-raw-escapes
Guido van Rossum added the comment:
I've submitted socket2.diff plus small changes to ssl.py.
This seems the best I can do given that I don't think I can make dup()
work on ssl sockets.
--
resolution: - accepted
status: open - closed
__
Tracker [EMAIL
billiejoex added the comment:
The current implementation of asynchat.async_chat.initiate_send method
doesn't look at what is specified in ac_out_buffer_size attribute which
represents the buffer size of the outgoing data defaulting to a maximum
of 4096 bytes to send in a single socket.send()
Changes by Brett Cannon:
--
assignee: nnorwitz - brett.cannon
_
Tracker [EMAIL PROTECTED]
http://bugs.python.org/issue1631171
_
___
Python-bugs-list mailing list
James G. sack (jim) added the comment:
I found the errror in my previous patch. It lacked a self.decode=.. line
in the StreamReader.decode elif branch.
I attach a replacement patch diff-u.py26_utf8sig (apply to the 2.6 version
of utf_8_sig.py. (If allowed, I will next remove the incorrect
67 matches
Mail list logo