Anthony Sottile added the comment:
> contain features we all want to get rid of
I don't think even this is unanimous. Things like registering codecs,
instrumenting coverage in subprocesses, etc. all seem like legitimate uses of
the arbitrary code execution
Anthony Sottile added the comment:
> What I think Anthony is looking for are ways to register “start up functions”
> that get executed automatically when the Python interpreter starts up
yes, this is what I want to still exist :)
my hope is that there's a clear standards-track
Anthony Sottile added the comment:
I don't have time to look through the data today but I wrote a script to
collect the usages of `.pth` from pypi. I realized after I ran it that I
skipped source distributions with `.zip` extension but otherwise it's pretty
complete:
https://
Anthony Sottile added the comment:
I did my best to classify those on pypi that were using `.pth` files. My
initial search had quite a few false positives (and now that I look at it,
completely missed `.zip`-based source distributions so there's likely some
false negatives as well)
H
Anthony Sottile added the comment:
> > There was a single .pth file that I deemed "malicious" since it
completely breaks the `subprocess` module (`subprocess-run`)
>
> It only seems to set an attribute. What's wrong with that? Does the early
import of subprocess
Anthony Sottile added the comment:
> Doesn’t that kind of prove my point? :)
It's not any worse than gevent ~breaking~ monkeypatching almost the entire
standard library. And to be fair to the author, it was created well before
(2013-06-21) python3.5's `run` api existed (2015
Anthony Sottile added the comment:
I think nearly all of the use cases in the packages are valid (except
module-layout) -- or at least if this feature were removed without having a
startup-time site-packages code execution feature there would be no possible
replacement. I'll elabor
Anthony Sottile added the comment:
> What I'm dismissing is that "pip install some-package" can define a global
> startup task for your interpreter. I shouldn't get debugging or code coverage
> enabled every time I run "python" just because I installed
Anthony Sottile added the comment:
>>> I should have to start that package somehow.
>>
>> `pip install` is a pretty good opt-in already imo
>
> Except that it conflates responsibilities. I may not want to opt into
> coverage even being loaded in my application b
New submission from Anthony Sottile :
The current code for `os.path.expanduser` in `ntpath` uses `HOME` in preference
to `USERPROFILE` / `HOMEDRIVE\\HOMEPATH`
I can't find any documentation of `HOME` being relevant on windows (only on
posix). For example, wikipedia only men
Change by Anthony Sottile :
--
keywords: +patch
pull_requests: +12262
stage: -> patch review
___
Python tracker
<https://bugs.python.org/issue36264>
___
_
Anthony Sottile added the comment:
> Modifying this value could have a real impact on the rest of the process, so
> we should be very careful to undo it regardless of test result. (Modifying
> HOME is not a as big a deal since, as you point out, it's not "real" ;) )
New submission from Anthony Sottile :
While building python 3.5.7 for https://github.com/deadsnakes
../Modules/_pickle.c: In function 'PyMemoTable_Copy':
../Modules/_pickle.c:677:5: error: 'for' loop initial declarations are only
allowed in C99 mode
for (si
Change by Anthony Sottile :
--
keywords: +patch
pull_requests: +12556
stage: -> patch review
___
Python tracker
<https://bugs.python.org/issue36478>
___
_
Anthony Sottile added the comment:
Here's an example in the wild which still reproduces with python3.8a3:
https://github.com/SecureAuthCorp/impacket/blob/194b22ed2fc85c4f241375fb7ebe4e0d89626c8c/impacket/examples/remcomsvc.py#L1669
This was reported as a bug on flake8:
https://gitla
Anthony Sottile added the comment:
It's a rather oldish gcc in this case. (4.8.4 + whatever ubuntu patches).
Here's a full (successful) build log (including the combinations of flags)
after patching:
https://launchpadlibrarian.net/416985438/buildlog_ubuntu-trusty-amd64.python3
Anthony Sottile added the comment:
Should this also produce warnings for `list` / `dict` / `set` literals?
```
$ python3.8
Python 3.8.0a3 (default, Mar 27 2019, 03:46:44)
[GCC 7.3.0] on linux
Type "help", "copyright", "credits" or "license" for more
New submission from Anthony Sottile :
This appears to be a regression between 3.8a3 and 3.8a4 -- though it may be
intentional and I'm missing something?
I noticed this while packaging 3.8 for deadsnakes
https://github.com/deadsnakes/python3.8
I've created a minimal reproductio
Anthony Sottile added the comment:
+vstinner
Hmmm, actually the relevant commit appears to be
6c44fde3e03079e0c69f823dafbe04af50b5bd0d and intentional
(I ran a git bisect to find this!)
https://github.com/python/cpython/pull/12931
Does PEP 3149 require an update with this change, it still
Anthony Sottile added the comment:
seems intentional => closing
--
stage: -> resolved
status: open -> closed
___
Python tracker
<https://bugs.python.or
New submission from Anthony Sottile :
More fallout from the Constant change in https://bugs.python.org/issue32892
minimal reproduction:
import ast
class V(ast.NodeVisitor):
def visit_Str(self, node):
print(node.s)
def main():
V().visit(ast.parse('x = "hi"&#
Anthony Sottile added the comment:
wrong bpo, this is the right one: https://bugs.python.org/issue32892
--
___
Python tracker
<https://bugs.python.org/issue36
Anthony Sottile added the comment:
hitting this in https://bugs.python.org/issue36917?
Is the simplification here really worth the breaking change to consumers?
I now have to write something that's essentially this to work around this which
feels more like the complexity has just
Anthony Sottile added the comment:
There would still be a breakage for that if someone was defining py36+
`visit_Constant` (which would clobber the `ast.NodeVisitor.visit_Constant` if
we were to add it)
--
___
Python tracker
<ht
Anthony Sottile added the comment:
spent some more time thinking about this and I think we should strongly
consider reverting. simplification in the core interpreter should not be
weighed lightly against complexity and breaking changes for users.
the change is also unfortunate because it
Anthony Sottile added the comment:
The simplest case is just the addition of an `isinstance` check:
https://github.com/asottile/dead/blob/85f5edbb84b5e118beab4be3346a630e41418a02/dead.py#L165-L170
class V(ast.NodeVisitor):
def visit_Str(self, node):
...
def visit_Constant
Anthony Sottile added the comment:
> You can not use the same implementation of the visitor for Num, Str,
> NameConstant and Ellipsis, because all these classes use different attribute
> for saving the value
ah yes, this is true -- maybe the better change would be to just add `
New submission from Anthony Sottile :
On windows, a deep path can be accessed by prefixing the path with \\?\
https://msdn.microsoft.com/en-us/library/windows/desktop/aa365247(v=vs.85).aspx?f=255&MSPPError=-2147217396#maxpath
The call to `listdir()` fails because it uses a posix separator
Change by Anthony Sottile :
--
keywords: +patch
pull_requests: +5023
stage: -> patch review
___
Python tracker
<https://bugs.python.org/issue32539>
___
___
Py
Anthony Sottile added the comment:
Seeing this as well when running the cpython test suite in docker:
```
$ ./python -m test.test_asyncio
...
[18 similar traces omitted]
==
ERROR: test_sock_sendfile_zero_size
Anthony Sottile added the comment:
Actually, my issue seems to be something more strange.
The host being passed in is `localhost` which resolves to:
```
>>> pprint.pprint(socket.getaddrinfo('localhost', 80))
[(,
,
6,
'',
('127.0.0.1', 80)),
Anthony Sottile added the comment:
Applying this patch makes the tests pass for me, but I don't think the patch is
appropriate (just hides the bug):
```
$ git diff
diff --git a/Lib/test/test_asyncio/test_unix_events.py
b/Lib/test/test_asyncio/test_unix_events.py
index 5bd76d3..ff6c4e1 1
Anthony Sottile added the comment:
Still a problem in 3.7:
$ python3.7
Python 3.7.0b2 (default, Feb 28 2018, 06:59:18)
[GCC 5.4.0 20160609] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import ast
&g
Change by Anthony Sottile :
--
pull_requests: +5909
___
Python tracker
<https://bugs.python.org/issue31550>
___
___
Python-bugs-list mailing list
Unsubscribe:
Anthony Sottile added the comment:
I made a new PR which instead *reverts* the python2.7 patch to restore
consistency
--
___
Python tracker
<https://bugs.python.org/issue31
Anthony Sottile added the comment:
I think the main concern here is ease in portability coupled with the
incorrectness of the current message (pointed out in
https://bugs.python.org/issue31550#msg302738)
For instance it was consistent in 2.7.1, but not later on in the 2.7.x tree.
*And
Anthony Sottile added the comment:
The intention of the change in issue 26510 was to pick the least surprising
behaviour for the default value of subparsers -- the compatiblity with the
behaviour before the regression was introduced in 3.3 was a nice side-effect.
As with the rest of
Anthony Sottile added the comment:
Grabbed the wrong SO link, I believe this is the one I meant to link to:
https://stackoverflow.com/a/18283730/812183
--
___
Python tracker
<https://bugs.python.org/issue33
Anthony Sottile added the comment:
Yeah, I picked the default value `True` because I couldn't actually find a user
of subparsers that _wanted_ optional subparsers. ¯\_(ツ)_/¯
--
___
Python tracker
<https://bugs.python.org/is
Anthony Sottile added the comment:
According to the other bugs, the change in 3.3 was an inadvertent regression.
The fact that it didn't get fixed for so long is mostly due to the unmaintained
state of argparse in the stdlib. The change in behaviour here is the _fix_ of
that regre
Anthony Sottile added the comment:
Shouldn't you wait for Raymond's arguments before outright closing the PR?
--
___
Python tracker
<https://bugs.python.o
New submission from Anthony Sottile :
Fortunately, this can be reproduced with the testsuite:
```
==
ERROR: test_copystat_symlinks (__main__.TestShutil
Anthony Sottile added the comment:
Here's one idea for a patch (inspired by the rest of the function):
```
diff --git a/Lib/shutil.py b/Lib/shutil.py
index 464ee91..2099289 100644
--- a/Lib/shutil.py
+++ b/Lib/shutil.py
@@ -213,6 +213,13 @@ def copystat(src, dst, *, follow_symlinks
Change by Anthony Sottile :
--
keywords: +patch
pull_requests: +4230
stage: -> patch review
___
Python tracker
<https://bugs.python.org/issue31940>
___
___
Py
Change by Anthony Sottile :
--
pull_requests: +4684
___
Python tracker
<https://bugs.python.org/issue31940>
___
___
Python-bugs-list mailing list
Unsubscribe:
New submission from Anthony Sottile :
While investigating https://bugs.python.org/issue31940 I noticed the following
is raised as `SystemError` instead of the expected `NotImplementedError`
(note: you need a platform with fchmodat but does not support nofollow)
```
touch foo
ln -s foo bar
Change by Anthony Sottile :
--
keywords: +patch
pull_requests: +4696
stage: -> patch review
___
Python tracker
<https://bugs.python.org/issue32277>
___
___
Py
Anthony Sottile added the comment:
I noticed it when I changed the test preconditions in
https://github.com/python/cpython/pull/4783
I tried changing a test to trigger this (in this branch) but I found I was just
implementing exactly the test in the `skip` condition which to me didn't
Anthony Sottile added the comment:
if I'm reading the manpage correctly: `readlink` tells the filename that the
symlink points to. lchmod is concerned with setting the `stat` on the link
itself (which only some platforms actually su
Anthony Sottile added the comment:
Should this have landed in python3.6? It removes a public symbol
`PyExc_RecursionErrorInst` (abi break?)
--
nosy: +Anthony Sottile
___
Python tracker
<https://bugs.python.org/issue30
Anthony Sottile added the comment:
Considering the huge popularity of these SO questions, I don't think this
should be reverted:
- https://stackoverflow.com/questions/23349349/argparse-with-required-subparser
-
https://stackoverflow.com/questions/22990977/why-does-this-argparse-code-b
Anthony Sottile added the comment:
Is there then no pathway for actually fixing the bug? aka how can I get
`required=True` to be the default.
--
___
Python tracker
<https://bugs.python.org/issue33
Anthony Sottile added the comment:
That's a separate issue (also a bug introduced by the bad 3.3 patch):
https://bugs.python.org/issue29298
I have an open PR to fix it as well but it has not seen review action:
https://github.com/python/cpython/pull
Anthony Sottile added the comment:
The bug is orthogonal, you can trigger it without the `required=` keyword
argument via the (currently suggested) monkeypatch workaround which restores
the pre-3.3 behaviour:
import argparse
parser = argparse.ArgumentParser()
subp = parser.add_subparsers
Change by Anthony Sottile :
--
pull_requests: +6992
___
Python tracker
<https://bugs.python.org/issue17909>
___
___
Python-bugs-list mailing list
Unsubscribe:
Anthony Sottile added the comment:
Unclear if this regression (from this patch) is intentional or not:
```
$ python3.6 -c 'import configparser; configparser.ConfigParser(defaults={"a":
None})'
$ python3.7 -c 'import configparser; configparser.ConfigParser(defaults
Change by Anthony Sottile :
--
keywords: +patch
pull_requests: +8005
stage: -> patch review
___
Python tracker
<https://bugs.python.org/issue19891>
___
___
Py
Anthony Sottile added the comment:
I was able to reproduce both the `PermissionError` and the `FileNotFoundError`
under these circumstances:
$ docker run --user 123:123 -ti python python
Python 3.7.0 (default, Jul 17 2018, 11:04:33)
[GCC 6.3.0 20170516] on linux
Type "help",
Change by Anthony Sottile :
--
nosy: +Anthony Sottile
___
Python tracker
<https://bugs.python.org/issue12782>
___
___
Python-bugs-list mailing list
Unsubscribe:
Anthony Sottile added the comment:
It appears it has always had this bug since introduction of absolute/relative
imports in
https://github.com/python/cpython/commit/f7f438ba3b05eb4356e7511401686b07d9dfb6d8
Agree with changing this to `# type: int` and correcting the documentation
Anthony Sottile added the comment:
In fact, trying to use an `ImportFrom` without an integer `level` results in a
`ValueError`:
>>> x = ast.parse('from os import path')
>>> x.body[0].level = None
>>> compile(x, '', 'exec')
Traceback (
Anthony Sottile added the comment:
Here's a simpler reproduction without involving a third party library:
>>> import cProfile
>>> from datetime import timedelta
>>> pr = cProfile.Profile()
>>> timedelta.total_seconds(-25200)
Traceback (most recent cal
Anthony Sottile added the comment:
Hmmm, I don't think mypy has an annotation for "sometimes has an attribute" --
`Optional[T]` is `Union[T, None]` (why I tried `None`).
But you're right, `FromImport` is constructable without a `level` -- it seems
to behave as `level=0`
Change by Anthony Sottile :
--
pull_requests: +9361
___
Python tracker
<https://bugs.python.org/issue16806>
___
___
Python-bugs-list mailing list
Unsubscribe:
Anthony Sottile added the comment:
This change in behaviour is breaking pycodestyle:
https://github.com/PyCQA/pycodestyle/issues/786
Perhaps it shouldn't have been backported (especially all the way to python2.7?)
--
nosy: +Anthony So
Anthony Sottile added the comment:
I'm surprised this was classified as a bug! Though that's subjective so I get
that it's difficult to decide what is and what isn't ¯\(ツ)/¯
--
___
Python tracker
<https://bug
Change by Anthony Sottile :
--
nosy: +Anthony Sottile
___
Python tracker
<https://bugs.python.org/issue33944>
___
___
Python-bugs-list mailing list
Unsubscribe:
Change by Anthony Sottile :
--
keywords: +patch
pull_requests: +9494
stage: needs patch -> patch review
___
Python tracker
<https://bugs.python.org/issu
Anthony Sottile added the comment:
Does this actually make sense for the `os` module? `PWD` is a variable set by
your interactive shell and doesn't really make sense outside that context.
I expect this function to be too easily confused with `os.getcwd()` and a
source of bugs whe
New submission from Anthony Sottile :
This may be intentional, but the behaviour changed between python2 and python3.
Want to make sure it's intentional as we're working (hacking) around this in
pytest: https://github.com/pytest-dev/pytest/pull/4284
The actual impact on pytest is
Anthony Sottile added the comment:
arbitrary, sure, but deriving from `Exception` maybe?
--
___
Python tracker
<https://bugs.python.org/issue35137>
___
___
Pytho
Anthony Sottile added the comment:
I'm seeing this same failure in python3.5 on 16.04 about 20% of the time:
```
$ python3.5 -m test -v test_dbm
== CPython 3.5.2 (default, Nov 17 2016, 17:05:23) [GCC 5.4.0 20160609]
== Linux-4.4.0-57-generic-x86_64-with-Ubuntu-16.04-xenial little-e
Anthony Sottile added the comment:
Stepping through the code, it seems under ndbm it is creating a file with a
'.db' extension:
```
(Pdb) list
47 'g': b'intended',
48 }
49
50 def init_db(self):
51 i
Anthony Sottile added the comment:
That doesn't seem to be the problem though, that occurs in both the successful
and failure case
--
___
Python tracker
<http://bugs.python.org/is
Changes by Anthony Sottile :
--
nosy: +Anthony Sottile
___
Python tracker
<http://bugs.python.org/issue28700>
___
___
Python-bugs-list mailing list
Unsubscribe:
New submission from Anthony Sottile:
PEP420 makes __init__.py files optional:
https://docs.python.org/3/whatsnew/3.3.html#pep-420-implicit-namespace-packages
Though it seems without them, pkgutil.walk_packages does not function as
desired: https://docs.python.org/3/library/pkgutil.html
Changes by Anthony Sottile :
--
nosy: +asottile
___
Python tracker
<http://bugs.python.org/issue16806>
___
___
Python-bugs-list mailing list
Unsubscribe:
Anthony Sottile added the comment:
Any updates on this? I'm running into this as well (still a problem in 3.4)
```$ python3.4
Python 3.4.2 (default, Oct 11 2014, 17:59:27)
[GCC 4.4.3] on linux
Type "help", "copyright", "credits" or "license" for m
New submission from Anthony Sottile:
First some expected output:
```
# from cmd.exe
C:\Users\Anthony>echo hi{1}
hi{1}
# from MINGW
$ echo hi{1}
hi{1}
```
```
# On ubuntu
$ echo 'hi{1}'
hi{1}
$ python3.5 -c "import subprocess; print(subprocess.check_output(('echo&#
Anthony Sottile added the comment:
It *is* in my path (otherwise it wouldn't produce any output at all). I'm not
trying to use the shell builtin, I'm trying to use the executable.
--
resolution: not a bug ->
status: closed -> open
_
Anthony Sottile added the comment:
To clarify further, the echo.exe on my path reacts correctly:
```
Anthony@AnthonysDesktop MINGW64 ~/Desktop/git/pre-commit
(allow_curly_braces_in_args)
$ /usr/bin/echo hi{1}
hi{1}
Anthony@AnthonysDesktop MINGW64 ~/Desktop/git/pre-commit
Anthony Sottile added the comment:
```
C:\Users\Anthony> C:\Users\Anthony\AppData\Local\Programs\Git\usr\bin\echo.exe
hi{1}
hi1
```
Must be the provider of echo.exe. I'll take it up with them
Sorry for the trouble!
--
resolution: -> not a bug
status: ope
New submission from Anthony Sottile:
I've confirmed this bug is present on both windows and linux, the outputs below
are from linux however.
Compare:
```
$ python3.4 --version
Python 3.4.3
$ python3.4 -c 'import os; print(os.unlink == os.remove)'
True
```
```
$ python3.5 -
Anthony Sottile added the comment:
Breaks this function:
```
def rmtree(path):
"""On windows, rmtree fails for readonly dirs."""
def handle_remove_readonly(func, path, exc): # pragma: no cover (windows)
excvalue = exc[1]
if fun
Anthony Sottile added the comment:
When calling shutil.rmtree on windows on a readonly directory, the error
handler is called with os.unlink as the first argument `func` which fails the
check `func in (os.rmdir, os.remove)` which succeeded on previous python
versions
Anthony Sottile added the comment:
Here's the workaround I'm opting for:
if sys.platform =='win32':
distutils.spawn.find_executable(cmd[0]) + cmd[1:]
--
nosy: +Anthony Sottile
___
Python tracker
<http://bu
New submission from Anthony Sottile:
Originally from https://github.com/testing-cabal/mock/issues/350
## Example
```python
from unittest import mock
class C(object):
def f(self):
pass
c = C()
with mock.patch.object(c, 'f', autospec=True):
with mock.patch.ob
Anthony Sottile added the comment:
The root cause seems to be that autospecced functions return a function object
(not a Mock instance) which a '.mock' attribute which is a MagicMock ( assigned
here:
https://github.com/python/cpython/blob/ae775ab1eb72f42de2d070158bade4bf261ac04f/Li
Anthony Sottile added the comment:
Here's an improved patch which:
- passes the tests
- puts the test in the correct place
I'm not entirely happy with the approach -- open to suggestions :)
--
Added file: http://bugs.python.org/file42
New submission from Anthony Sottile:
Patch attached with test.
In summary:
A request to the url b'/\x80' appears to the application as a request to
b'\xc2\x80' -- The issue being the latin1 decoded PATH_INFO is re-encoded as
UTF-8 and then decoded as latin1
(on the wi
Anthony Sottile added the comment:
A few typos in my previous comment, pressed enter too quickly, here's an
updated comment:
Patch attached with test.
In summary:
A request to the url b'/\x80' appears to the application as a request to
b'/\xc2\x80' -- The issue
Anthony Sottile added the comment:
Oops, broke b'/%80'.
Here's a better fix that now takes:
(on the wire) b'\x80' -(decode latin1)-> u'\x80' -(encode utf-8)->
b'\xc2\x80' -(decode latin1)-> u'\xc2\x80'
to:
(on the wire
Anthony Sottile added the comment:
Updates after review.
--
Added file: http://bugs.python.org/file42404/patch
___
Python tracker
<http://bugs.python.org/issue26
Anthony Sottile added the comment:
Forgot to remove the pyver code (leaning a bit too much on pre-commit)
--
Added file: http://bugs.python.org/file42405/patch
___
Python tracker
<http://bugs.python.org/issue26
Anthony Sottile added the comment:
PEP states that environ variables are str variables decoded using
latin1:
https://www.python.org/dev/peps/pep-/#id19
Therefore, to get the original bytes, one must encode using latin1
On Apr 20, 2016 3:46 AM, "Александр Эри" wrote:
>
&g
Anthony Sottile added the comment:
I have a hunch that this fix here may be causing this:
https://github.com/spotify/dh-virtualenv/issues/148
Minimally:
echo 'from setuptools import setup; setup(name="demo")' > setup.py
echo 'import pytest' > tests/__init_
Anthony Sottile added the comment:
Seems I've named the patchfile incorrectly -- Hopefully this is correct this
time?
--
keywords: +patch
nosy: +Anthony Sottile
Added file: http://bugs.python.org/file42675/2.patch
___
Python tracker
Anthony Sottile added the comment:
Adding `import gc; gc.collect()` doesn't change the outcome afaict
--
nosy: +asottile
___
Python tracker
<http://bugs.python.org/is
Anthony Sottile added the comment:
I'm still seeing a very large difference:
asottile@work:/tmp$ python repro.py
ready
72604
VmHWM: 72604 kB
VmRSS: 60900 kB
asottile@work:/tmp$ rm *.pyc; python repro.py
ready
1077232
VmHWM: 1077232 kB
VmRSS:218040 kB
This fi
Anthony Sottile added the comment:
3.4 seems happier:
asottile@work:/tmp$ rm *.pyc; python3.4 repro.py
ready
77472
VmHWM: 77472 kB
VmRSS: 65228 kB
asottile@work:/tmp$ python3.4 repro.py
ready
77472
VmHWM: 77472 kB
VmRSS: 65232 kB
The nasty result above is from 2.7:
$ python
301 - 400 of 403 matches
Mail list logo
