Mailman 3 October 2016 - Distutils-SIG

Commit 17f89f4 "Update sdist to use sdist_add_defaults forward compatibility"
by Brandon Casey 21 Oct '16

21 Oct '16

Hi Jason, I just started looking at the setuptools repo last night to see if I could figure out why setuptools doesn't place data_files inside an sdist and I saw your commit 17f89f4 to teach the sdist command to use the python 36 compatibility behavior. But it overrides _add_defaults_data_files() which actually adds the data_files to the filelist, and replaces it with an empty function. If I remove this function, so that the one from py36compat.sdist_add_defaults is used, then it seems to work correctly for me. But that is the extent of my testing so far. The doc comment for the new _add_defaults_data_files function just says: Don't add any data files, but why? I see _add_defaults_python() was also overridden and is now only slightly different from the one in py36compat.sdist_add_defaults. It looks like maybe you were just trying to make use of the py36compat functions without introducing any change in behavior. Is that the only reason that _add_defaults_python() differs from the one in py36compat and why _add_defaults_data_files is disabled? Or do you have some concerns about enabling this functionality? Any thoughts of enabling it? If there's some active discussion or explanation on a mailing list somewhere, please feel free to point me towards that. Thanks, -Brandon

2 2

Interrogate distribution for an entry point command path
by Ben Finney 16 Oct '16

16 Oct '16

Howdy, How can a Python application discover at run-time where on the filesystem its own ‘entry_points’ programs are available? The Setuptools ‘entry_points’ are available at run-time to the distribution, via the ‘pkg_resources’ API for entry points <URL:https://setuptools.readthedocs.io/en/latest/pkg_resources.html#entry-points>. I don't see there how to interrogate for which filesystem path to invoke for a specific entry point's command. How can I reliably ask ‘pkg_resources’ for “where is the entry point named ‘foo’ installed?” such that I get a filesystem path of a command to invoke? -- \ “Geeks like to think that they can ignore politics. You can | `\ leave politics alone, but politics won't leave you alone.” | _o__) —Richard M. Stallman, 2002-07-26 | Ben Finney

6 10

manylinux vs 3.6.0
by Robin Becker 13 Oct '16

13 Oct '16

Any manylinux experts here? I tried to upgrade the docker used to build my linux packages, but I failed to get a build of python 3.6.0b2 1) I fixed a missing down load for openssl -OPENSSL_ROOT=openssl-1.0.2h -OPENSSL_HASH=1d4007e53aad94a5b2002fe045ee7bb0b3d98f1a47f8b2bc851dcd1c74332919 +OPENSSL_ROOT=openssl-1.0.2j +OPENSSL_HASH=e7aff292be21c259c6af26469c7a9b3ba26e9abaaffd325e3dccc9785256c431 2) I fixed the way the download of python source is done; seems the folder is always the numeric part of the python version and the a0 a1 b2 etc etc must be removed. - wget -q $PYTHON_DOWNLOAD_URL/$py_ver/Python-$py_ver.tgz + wget -q $PYTHON_DOWNLOAD_URL/${py_ver%[a-z][0-9]}/Python-$py_ver.tgz 3) The build of 3.6.0b2 started OK in the container, but failed with this message > + ./configure --prefix=/opt/_internal/cpython-3.6.0b2 --disable-shared > configure: WARNING: linux/random.h: present but cannot be compiled > configure: WARNING: linux/random.h: check for missing prerequisite headers? > configure: WARNING: linux/random.h: see the Autoconf documentation > configure: WARNING: linux/random.h: section "Present But Cannot Be Compiled" > configure: WARNING: linux/random.h: proceeding with the compiler's result > configure: WARNING: ## --------------------------------------- ## > configure: WARNING: ## Report this to https://bugs.python.org/ ## > configure: WARNING: ## --------------------------------------- ## > + make -j2 > Python/dtrace_stubs.o: In function `PyDTrace_LINE': > /Python-3.6.0b2/./Include/pydtrace.h:28: multiple definition of `PyDTrace_LINE' > Python/ceval.o:/Python-3.6.0b2/./Include/pydtrace.h:28: first defined here > Python/dtrace_stubs.o: In function `PyDTrace_FUNCTION_ENTRY': ...... > Modules/gcmodule.o: In function `PyDTrace_INSTANCE_DELETE_START_ENABLED': > /Python-3.6.0b2/./Include/pydtrace.h:45: multiple definition of `PyDTrace_INSTANCE_DELETE_START_ENABLED' > Python/ceval.o:/Python-3.6.0b2/./Include/pydtrace.h:45: first defined here > Modules/gcmodule.o: In function `PyDTrace_INSTANCE_DELETE_DONE_ENABLED': > /Python-3.6.0b2/./Include/pydtrace.h:46: multiple definition of `PyDTrace_INSTANCE_DELETE_DONE_ENABLED' > Python/ceval.o:/Python-3.6.0b2/./Include/pydtrace.h:46: first defined here > collect2: error: ld returned 1 exit status > make: *** [Programs/_freeze_importlib] Error 1 is the pydtrace stuff something new in the beta? I searched and found this issue https://bugs.python.org/issue28092 which seems related as it's the same centos version used in the dockerfile. As I understand it this means the new sources contain code which prevents compilation in the manylinux environment. -- Robin Becker

2 6

3.6.0b2 on windows
by Robin Becker 12 Oct '16

12 Oct '16

I have a question regarding an error message I see when building an extension for windows using the 3.6.0b2 release > C:\ux\XB33\py36_x86\lib\site-packages\wheel\pep425tags.py:77: RuntimeWarning: Config variable 'Py_DEBUG' is unset, Pytho > n ABI tag may be incorrect > warn=(impl == 'cp')): > C:\ux\XB33\py36_x86\lib\site-packages\wheel\pep425tags.py:81: RuntimeWarning: Config variable 'WITH_PYMALLOC' is unset, > Python ABI tag may be incorrect > warn=(impl == 'cp')): is this something I need to worry about? -- Robin Becker

2 1

Re: [Distutils] [Python-Dev] pythonhosted.org upload no longer works
by Paul Moore 11 Oct '16

11 Oct '16

On 10 October 2016 at 14:34, Giampaolo Rodola' <g.rodola(a)gmail.com> wrote: > This is what I've bumped into just now: > > python setup.py upload_sphinx --upload-dir=docs/_build/html > running upload_sphinx > Submitting documentation to https://upload.pypi.org/legacy/ > Upload failed (410): Uploading documentation is no longer supported, we > recommend using https://readthedocs.org/. > > Personally I prefer pythonhosted over readthedocs as I can provide the same > style as official Python's doc (see https://pythonhosted.org/psutil/) and, > most importantly, because I can automatize doc upload just by running "make > doc-upload". > Is there a reason upload functionality has been disabled? > Is pythonhosted.org going to be dismissed or something? This was discussed on distutils-sig back in May 2015. The thread starts here: https://mail.python.org/pipermail/distutils-sig/2015-May/026327.html. One of the actions in the final proposal (https://mail.python.org/pipermail/distutils-sig/2015-May/026381.html) included a step to contact projects that were using pythonhosted.org. Did you not get any contact? It might be worth picking this up on distutils-sig, as that's where the original discussion occurred. I've copied this reply to there, and suggest followups go to that list. Paul

5 5

pip enhancements: check for dependent packages before uninstalling and store installation date to allow listing by it
by João Matos 07 Oct '16

07 Oct '16

Hello, I believe it would be helpful if pip checked if there are any dependent packages before uninstalling a package. If there were, it should: 1. Warn the user by listing the dependent packages; 2. Require a specific option, eg. --alldeps to uninstall all dependent packages before uninstalling the specified package; 3. Require a specific option, eg. --force to continue with the uninstall w/o uninstalling the dependent packages (which could be dangerous, but it's the current behaviour and may be useful in specific situations). This is similar to what happens with package managers in Linux. This should also work on venvs. By storing the installation date of all packages, it should allow the list command to be ordered by it. Best regards, JM

1 0