New submission from Martin Panter:
https://docs.python.org/3.7/c-api/typeobj.html#c.PyTypeObject.tp_getattr
tp_getattr and tp_setattr take a non-const char pointer in their signatures,
but the documentation points to PyObject_GetAttrString() etc which were changed
to take const char pointers
Martin Panter added the comment:
I’m not sure I understand your questions Brett (Which tokenizer? What
rebuilding?), but I will try to explain some relevant parts.
My main goal is to add a makefile dependency of parsetok.o on $(GRAMMAR_H). In
Python 3, that is easy, and (I realize now) my
Martin Panter added the comment:
Hmm, it seems I already discovered this leak six months ago:
https://bugs.python.org/review/27100
--
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/i
Martin Panter added the comment:
https://mail.python.org/pipermail/python-checkins/2016-November/147171.html
test_collections leaked [0, 0, 7] memory blocks, sum=7
test_contextlib leaked [38, 38, 38] references, sum=114
test_contextlib leaked [13, 12, 12] memory blocks, sum=37
test_descr leaked
Martin Panter added the comment:
My Australian Style Manual says to use en rules for spans of figures etc.
Wikipedia mentions some other guides:
<https://en.wikipedia.org/wiki/Dash#Ranges_of_values>; apparently some prefer
the hyphen. That is why I said it is a style issue, not a
Martin Panter added the comment:
No opinion if this is a good change to make, but I left some review suggestions
--
nosy: +martin.panter
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/i
Martin Panter added the comment:
Sure, I was just saying that for whatever reason, 50% of the documentation uses
hyphens for number ranges, and 50% uses en dashes.
--
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/i
Martin Panter added the comment:
This was added to 2.7 in r68532 (Issue 4879), but the semicolon never made it
to the Python 3 branch that I can see, so I don’t think there is anything to
fix there. Your patch is against the 2.7 branch, which is only open for bug
fixes now, and I don’t think
Changes by Martin Panter <vadmium...@gmail.com>:
--
superseder: Warning -- warnings.filters was modified by test_warnings ->
Implement comparison (x==y and x!=y) for _sre.SRE_Pattern
___
Python tracker <rep...@bugs.pytho
Martin Panter added the comment:
As long as we are restricted by backwards compatibility, it will be hard to
find a hack-free solution. The ideal solution IMO is to re-create
_warnings.filters from scratch when _warnings is reloaded, but such a change
would be safer only for 3.7.
So I am
Martin Panter added the comment:
Personally I prefer en dashes where practical, but I fear the change may
conflict with other people’s styles and preferences. I quickly counted 22
single hyphens (obviously I missed a few that your patch caught) and also 22
existing en dashes
Martin Panter added the comment:
Thanks Raymond, now the test is noticeably faster and fine with my memory
situation :)
--
stage: -> resolved
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python.or
Martin Panter added the comment:
The committed change looks wrong in the new example. I think Mariatta’s patch
was right; it the index found is 3, not 2:
>>> a
[66.25, 333, -1, 333, 1, 1234.5, 333]
>>> a.index(333)
1
>>> a.index(333, 2) # search for 333 starting at
Martin Panter added the comment:
Gareth, are you sure that fixes the main memory problem? Did you see Serhiy’s
cum_weights list? Looking at the code, a list of every number from one to 13
million will use more memory (to hold each unique integer) than the initial
list of repeated ones.
I
Martin Panter added the comment:
The text looks okay. I’m not sure about putting it in the “Porting to Python
3.6” section, but there doesn’t seem to be a more appropriate existing heading.
Perhaps we need a new heading, say “Other Improvement”, near “Optimizations”
and “Build and C API
Martin Panter added the comment:
Ned indicated he was open to the idea:
<http://mid.mail-archive.com/nu39u2$9ri$1...@blaine.gmane.org>. Maybe just
double check; there’s still a few hours left!
I’m not sure, but maybe there should be “.. versionchanged:: 3.6” in the RST
documen
Martin Panter added the comment:
Equivalent patch for 2.7
--
Added file: http://bugs.python.org/file45576/graminit-dep.py2.patch
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/
Changes by Martin Panter <vadmium...@gmail.com>:
--
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
___
Python tracker <rep...@bugs.python.org>
<http://bu
Changes by Martin Panter <vadmium...@gmail.com>:
--
resolution: -> fixed
stage: commit review -> resolved
status: open -> closed
___
Python tracker <rep...@bugs.python.org>
<http://bu
Martin Blais added the comment:
Well that just echoes exactly what I originally thought, but somebody else said
it was not needed because ExitStack already exists and could be used for that
purpose.
If this were at work and/or it were all just to me, I'd just implement a brand
new
Martin Blais added the comment:
Adding nullcontext = ExitStack in the source file would solve this problem in a
single line of code.
--
nosy: +blais
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/i
Martin Panter added the comment:
Okay I will assume that $(srcdir) reference was just due to a mistake or bad
cargo-culting then.
Will commit this soon unless anyone has a good idea to avoid embedding
@abs_srcdir@.
--
stage: patch review -> commit rev
Martin Panter added the comment:
Okay committed to 3.7 for the moment. I think that is all we can reasonably do
unless we drop the pseudo-HTTP-0.9 support.
--
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
_
Martin Panter added the comment:
Thanks Mariatta. Some people like to use the warn(stacklevel=2) parameter, then
the warning message is potentially more useful.
Also, I think it would be good to write a brief test case for the warning.
There are probably lots of examples in the test suite
Changes by Martin Panter <vadmium...@gmail.com>:
--
stage: -> patch review
versions: +Python 3.6, Python 3.7
___
Python tracker <rep...@bugs.python.org>
<http://bugs.pyt
Martin Panter added the comment:
Here is a patch implementing Eryk’s suggestion, for both from_buffer() and
from_buffer_copy() methods. This is exactly how it is already handled for
Array.from_address(). The two buffer methods were added more recently than
from_address().
--
keywords
New submission from Martin Panter:
Revision 32bfc8b6 added
test.test_random.MersenneTwister_TestBasicOps.test_choices_algorithms(), which
runs the following code:
n = 13132817 # 13 million
self.gen.choices(range(n), [1]*n, k=1)
When I build Python with the “--with-pydebug” configure
Martin Panter added the comment:
Here is another way to remember that the filter list has already been
initialized. I made a new immortal _warnings.filters_initialized flag at the C
level. It is actually a list so that it can be mutated and remembered across
module reloads, but it is either
Changes by Martin Panter <vadmium...@gmail.com>:
Added file: http://bugs.python.org/file45542/immortal-filters.patch
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python
Martin Blais added the comment:
I've been looking for this today; I would have used it.
Instead of an obvious (and explicit) null context manager, I had to read
through this entire thread to eventually find out that I can use something
called ExitStack(), which is designed for another use
Martin Panter added the comment:
Having an unassigned f-string as the first statement is still valid syntax, so
could at most be a warning, not a SyntaxError.
--
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/i
Martin Panter added the comment:
Actually, testing your code fragment, it seems you do get a doc string when the
f-string has no substitutions in curly brackets, otherwise you don’t get any
doc string. Maybe this is due to how different forms of string are compiled.
>>>
Martin Panter added the comment:
There was a bit of discussion at the top of
<https://bugs.python.org/issue25179#msg254298>.
IMO it would be simpler do disallow all f-strings as docstrings. Otherwise,
what would the result of this be:
name = "module level"
class C:
nam
Martin Panter added the comment:
I think other people wanted to add an automated deprecation warning in the code
first, before removing it.
--
nosy: +martin.panter
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/i
Changes by Martin Panter <vadmium...@gmail.com>:
Removed file: http://bugs.python.org/file42579/downloadfile.htm
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python
Martin Panter added the comment:
I didn’t really like adding the _add_filter() special handling in the first
place, but I went ahead because I couldn’t think of a better way to avoid the
problem with reloading the warnings modules. So unless anyone can suggest
anything better, I am okay
Martin Panter added the comment:
That’s not exactly a duplicate; one is host-by-NAME, the other host-by-ADDR
--
nosy: +martin.panter
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/i
Martin Panter added the comment:
Is the problem something like a missing C function prototype? Maybe you see
compiler warnings, but the compiler and linker carry on with the wrong
prototype. If you build with “make -s”, warnings might be easier to see.
If my guess is right, this would
Martin Panter added the comment:
BTW my understanding is that currently, if the garbage collector deletes a
generator, it effectively calls its close() method. When close() is called, it
throws GeneratorExit into the suspended generator. This is meant to cause all
the “with” and “try
Martin Panter added the comment:
Perhaps I shouldn’t have mentioned “del gen”. That was meant to represent the
garbage collector running and calling gen.__del__() or equivalent. Basically
what I was trying to say is I think there will be two classes of generators:
1. Simple generators just
Martin Panter added the comment:
I committed your patches for 3.5+ and 2.7.
--
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
versions: +Python 3.7
___
Python tracker <rep...@bugs.python.org>
<http
Changes by Martin Panter <vadmium...@gmail.com>:
--
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
versions: +Python 2.7, Python 3.7
___
Python tracker <rep...@bugs.python.org>
<http://bu
Martin Panter added the comment:
This would be hard to get right, but maybe not impossible. If the frame is
suspended, and will do special handling of GeneratorExit, a ResourceWarning
would be nice in most cases. But if the frame will not catch any exception,
there should
Martin Panter added the comment:
In general I agree with the doc strings giving the main details, and leaving
smaller details for the reference documentation. Maybe for concrete
implementations like BytesIO, it is not worth saying the expanded contents are
undefined. One other way to make
Martin Panter added the comment:
When I run the tests with -Werror (or any other -W option), I now get
2 tests altered the execution environment:
test___all__ test_warnings
--
nosy: +martin.panter
___
Python tracker <rep...@bugs.python.
Martin Panter added the comment:
Thanks Lewis
--
resolution: -> fixed
stage: needs patch -> resolved
status: open -> closed
___
Python tracker <rep...@bugs.python.org>
<http://bugs.pyt
Martin Panter added the comment:
The offending commit is revision 6b70f16d585a. The true name is numeric_owner;
numeric_only is wrong. The name also needs fixing in Doc/whatsnew/3.5.rst.
--
nosy: +martin.panter
___
Python tracker <
Martin Panter added the comment:
Issue 11427 has some discussion about this. It seems one concern was to make it
hard to modify the buffer of a bytes object (which is supposed to be
immutable). But I agree there should be an easy way to access read-only buffers
in ctypes. Perhaps that could
Martin Wagner added the comment:
i have a use-case that requires a behavior that is referenced above as
--canonicalize-missing. essentially i need to get rid of relative parts in a
path representation regardless the actual filesystem.
my conclusion was that PurePath could provide
Changes by Martin Panter <vadmium...@gmail.com>:
--
resolution: -> fixed
stage: commit review -> resolved
status: open -> closed
___
Python tracker <rep...@bugs.python.org>
<http://bu
Martin Panter added the comment:
Here is a patch for Python 3 implementing my idea. There is already code
conditionally compiled out for the pgen build, so my patch just expands that to
avoid the "graminit.h" include and derived functions.
--
stage: needs patch -> patch
Martin Panter added the comment:
My patch doesn’t help answer the question of why this only fails on BSD, so I
will hold off on committing it. However v2 has a minor update to skip the test
when pthread_sigmask() is unavailable.
--
versions: +Python 3.7
Added file: http
Martin Panter added the comment:
Actually I think it is good to document DESTDIR (you weren’t seem aware of it
before?). I was just pointing out that it is useful to more than just Android.
Perhaps Guido shouldn’t have trimmed it from the 2.7 readme (original text:
https://hg.python.org
Martin Panter added the comment:
I haven’t looked at your code, but I think you may have misunderstood how
variable assignments work in Python. See
<https://docs.python.org/3.5/faq/programming.html#why-did-changing-list-y-also-change-list-x>.
When you assign an object such as
Changes by Martin Panter <vadmium...@gmail.com>:
--
versions: +Python 3.7 -Python 3.5
Added file: http://bugs.python.org/file45300/parse-version.v2.patch
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python
Martin Panter added the comment:
Regarding 2.7, I guess it depends on your definition of “support”. People have
been cross-compiling Python 2.7 longer than I have been working on Python. In
the past I have tried to apply cross compilation fixes to 2.7. But if you are
only confident about 3.6
Martin Panter added the comment:
Consuming and ignoring 1xx responses seems reasonable for general cases.
Treating 101 (Switching Protocols) as a special case also seems reasonable.
I also agree it would be good to provide an API to return these responses (at
least after the request
Martin Panter added the comment:
Apparently (haven’t tried myself) if you put “.. program:: zipfile” before the
:option: invocations, it changes the default context and you don’t need the
bit.
The text in general looks fine. One very minor thing: I would use a hyphen when
“command line
Martin Panter added the comment:
Sorry, I should have said: the other two files (Misc/python.man and
Modules/main.c) also need an extra “the”, like you changed in
Doc/using/cmdline.rst.
Other than that, it looks okay to me. But I guess you could add a quick test
case in Lib/test
Martin Panter added the comment:
For the “unittest” module, I added separate text with a link: (see Warning
control). Similarly for “timeit” in 2.7. Now the only Sphinx warnings I get are
a handful about syntax highlighting, and two genuine ones relating to the
missing -b documentation (Issue
Martin Panter added the comment:
I left some review comments. Also (as a native English speaker), I think it is
okay to drop the third line about “multiple times”. You already say it may be
given twice in the line above.
--
nosy: +martin.panter
Martin Panter added the comment:
As well as the usage string, it is missing from the list of options in the main
documentation at Doc/using/cmdline.rst. There are a couple of places (sys.rst
and warnings.rst) that reference :option:`-b`, and newer versions of Sphinx
complain about
Martin Panter added the comment:
Clément: Yes, that sounds like the intended behaviour, at least when using
Readline. Originally discussed in Issue 23441 and Issue 22086. (Though I think
the change in behaviour when completion is called outside of Readline is a bug
Martin Panter added the comment:
Looks good to me :)
--
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/issue28444>
___
___
Python-bugs-
Martin Panter added the comment:
Thanks, this looks like a decent start. I’m not that familiar with
cross-compiling Python, but here are a couple questions:
$(cd /path/to/source && pwd)/configure
Why do you use $(pwd) here? I have mainly seen relative paths used (my own
exp
Martin Panter added the comment:
In general, I would be against stripping square brackets if the host and port
have already been separated, e.g. in the address tuple used with
socket.create_connection() etc. It feels like unnecessary double processing,
which should already have been done
Martin Panter added the comment:
Here is a patch to parse the version from the request in more cases. Since it
is more of a cosmetic improvement for handling erroneous requests, I would
probably only apply it to the next Python version (3.7 atm). Or do you think it
should go into bug fix
Martin Panter added the comment:
Here is a slightly updated patch I have been sitting on. I think the main
difference was tweaks to the documentation.
Issue 28548 would also benefit if the HTTP 0.9 code was removed. The presence
of the HTTP 0.9 code means a suboptimal response is sent in some
Martin Panter added the comment:
Since my last comment, I have become more confident that Python’s request
parsing never supported proper HTTP 0.9. So I would prefer to remove it in the
next version of Python, and not bother fixing Xiang’s bug in existing versions
(unless someone offers
Martin Panter added the comment:
Parsing the version from words[-1] rather than words[2] may be a minor
improvement however.
--
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/i
Martin Panter added the comment:
I think you should be able to reproduce this without Mailman or tox, by just
running “python -m http.server”.
The problem is the “HTTP/0.9” protocol that Python is assuming does not include
a header section, so there is no place to put a 400 status code
Martin Turon added the comment:
If I bind to port 0, it works for both tx and rx. Sorry for the false alarm!
sock.bind(("monitor0", 0))
--
status: open -> closed
___
Python tracker <rep...@bugs.python.org>
<http://bugs
Martin Turon added the comment:
Just for clarity, the high level bug is that when binding to an interface using
AF_PACKET, transmissions work, but receive does not:
sock = socket.socket(socket.AF_PACKET, socket.SOCK_RAW, ETH_P_IEEE802154)
sock.bind(("monitor0", ETH_P_
New submission from Martin Turon:
When binding to AF_PACKET linux kernel sockets, the interface name is not
passed in when given -- it is always "". This causes problems, for example,
receiving packets to a "monitor0" interface doesn't work.
diff -r a6548e230ed6 Mo
Martin Panter added the comment:
I tried the original revision 433606e9546c code in Python 2.7 (which still
allows mixed tabs and spaces), and it behaves the same as the current version.
Thus I suspect revision 9e2b94a3b5dc did not change any behaviour.
What is your use case? Why can’t you
Martin Panter added the comment:
You say this is a regression. Can you point to a version that worked as you
want? The IPv6 URL handling you pointed out seems to have been added to 2.4b1,
and then backported to 2.3.5. I expect that earlier versions didn’t support
IPv6 URLs at all.
Also
Changes by Martin Panter <vadmium...@gmail.com>:
--
resolution: -> fixed
stage: commit review -> resolved
status: open -> closed
___
Python tracker <rep...@bugs.python.org>
<http://bu
Martin Panter added the comment:
I occasionally get the following error, due to Parser/parsetok.o being older
than Include/graminit.h.
./python -E -S -m sysconfig --generate-posix-vars ; if test $? -ne 0 ; then
echo "generate-posix-vars failed" ; rm -f ./pybuilddir.txt ; exit 1 ;
Martin Panter added the comment:
Regarding out-of-tree builds (Problem 3), see Issue 10656, which already has a
potential patch.
--
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/i
Martin Panter added the comment:
Actually, why do you remove $(srcdir) for Modules/python.exp? This code was
added in r88426; see <https://bugs.python.org/issue941346#msg128586>.
--
___
Python tracker <rep...@bugs.python.org>
<http://
Martin Panter added the comment:
The idea of the patches makes sense to me.
Embedding the name of whatever directory Python was built in into the installed
version of Modules/ld_so_aix.in does not seem very clean to me. Is there
another way of doing this? Perhaps a file copy or symlink
Martin Panter added the comment:
Michael F.: It sounds like you have three separate but related problems:
1. Confusion between LDSHARED and BLDSHARED referring to the in-source build
tree vs final installed files. I think this is what David and Michael H. were
originally trying to fix here
Martin Panter added the comment:
Perhaps it is okay to keep the documentation changes, but I think either the
library changes should be reverted or worked around where the modules are still
in use.
I normally run the tests with -Werror, and the failures I get are:
* test_ssl, test_smtplib
Martin Panter added the comment:
I agree it is good to explicitly document the __enter__() result, rather than
relying on assumptions and example code. The patch looks good to me.
I don’t understand the problem with pop_all() though. Is there still a problem
if we apply your patch
Martin Panter added the comment:
Perhaps you should merge your work with Issue 2786. With a very brief look, the
patches seem to take a similar approach.
--
nosy: +martin.panter
resolution: -> duplicate
superseder: -> Names in function call exception should have class
Martin Panter added the comment:
I altered the comment (looks like it was a copy from code forcing proxies to be
bypassed). Also, I didn’t port the second no_proxy fix to 2.7; it looks like
there is a different workaround there which is not affected by the environment.
--
resolution
Changes by Martin Panter <vadmium...@gmail.com>:
Added file: http://bugs.python.org/file45180/subprocess3-2.7.patch
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python
Martin Panter added the comment:
Here are corresponding patches for 3.5 and 2.7.
--
stage: patch review -> commit review
Added file: http://bugs.python.org/file45179/subprocess3-3.5.patch
___
Python tracker <rep...@bugs.python.org&
Changes by Martin Panter <vadmium...@gmail.com>:
--
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
___
Python tracker <rep...@bugs.python.org>
<http://bu
Martin Panter added the comment:
Yep, working well. Thanks Guido & Ivan.
--
stage: commit review -> resolved
___
Python tracker <rep...@bugs.python.org>
<http://bugs.pytho
Martin Panter added the comment:
I will this to Git Hub when I get a chance to get set up, if that helps though
:)
--
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/i
Martin Panter added the comment:
Modifying the API to accept a keyword argument is not worthwhile IMO. That
means modifying modules like http.client and zipfile (which use the wrong
keyword name), and user-defined implementations may become incompatible.
The question of documentation
Martin Panter added the comment:
The minus sign might have been deliberate at some stage, but I realize it is
more accessible etc if it was ASCII -1, so I will change that.
The idea of changing the signature is to avoid people thinking it accepts a
keyword argument. See e.g. Issue 25030
Martin Panter added the comment:
I just got this failure again today. I think I have seen it once or twice
before. Is the failure actually indicating a problem with Python, or is the
test just too strict?
It seems that it may be like a test ensuring that a random.randint(1, 100) is
never
Changes by Martin Panter <vadmium...@gmail.com>:
--
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
___
Python tracker <rep...@bugs.python.org>
<http://bu
Martin Panter added the comment:
V3 looks good to me
--
___
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/issue26240>
___
___
Python-bugs-
Changes by Martin Panter <vadmium...@gmail.com>:
--
title: Built-in module 'Time' does not enable functions if -Wno-error specified
in the build environment -> Built-in module 'time' does not enable functions if
-Werror specified in the build en
Martin Panter added the comment:
I also committed a similar but independent fix in Python 2.7 building
Modules/_sqlite/connection.c, caused by revision 649937bb8f1c, and adjusted
some tests to work when multithreading is disabled.
For the record, I also opened Issue 28482, Issue 28484
New submission from Martin Panter:
According to the documentation and tests, a negative value for the “workers”
parameter to compileall.compile_dir() should trigger ValueError. But if Python
is built with “configure --without-threads”, the parameter is not checked. So I
propose a patch to fix
New submission from Martin Panter:
Two tests in test_capi assume that the GIL is always used, but I don’t think it
is available when Python is built with “configure --without-threads”. So I
propose to skip these tests if the “threading” module is unavailable.
In test_regrtest, a test is rerun
901 - 1000 of 9608 matches
Mail list logo
