-
-
Notifications
You must be signed in to change notification settings - Fork 10.9k
Numpy as setuptools dependency (Trac #860) #1458
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Comments
Attachment added by trac user phsilva on 2008-07-24: patch-numpy-1.1.0 |
@charris wrote on 2008-07-24 There have been some modifications to this file and others for 1.1.1 (and trunk). Can you check if the problem persists? |
trac user phsilva wrote on 2008-07-24 Replying to [comment:1 charris]:
Problem persist in trunk and numpy 1.1.1rc2. |
trac user phsilva wrote on 2008-10-30 Guys, can you please take a look at this bug? its just one line patch and the "bug" persists in 1.2.1. Cheers, |
trac user ceball wrote on 2009-10-01 Just wondering if there's any news about this? numpy 1.3 also doesn't seem to work as a dependency:
|
trac user tdavis wrote on 2010-01-19 This issue remains as of numpy 1.4.0 ...
|
trac user phsilva wrote on 2010-01-19 C'mon guys, say WONTFIX or fix it, please. More than year and a half with this one line bug open and not a single statement about it. |
@charris wrote on 2010-01-19 Send a heads up to the mailing list. |
@cournape wrote on 2010-01-20 Sandbox issues in setuptools are complicated. It is also difficult to reproduce, and easy_install is deprecated anyway. |
trac user phsilva wrote on 2010-01-20 Thanks for the answer. Are you guys going to support Distribute and pip? I'm instructing people to just use whatever tool their distribution provides (apt, rpm) to install numpy anyway. setuptools was soo complicated with large packages that have custom distutils. |
@cournape wrote on 2010-01-20 I am afraid we will have to support Distribute, yes. pip should cause less issues than easy_install, because it uses sources and not eggs. If you have issue with pip, please report them. Using the distribution packages is certainly a good advice - distutils/setuptools is a clusterfuck, and as long as a better solution is not adopted, those issues will persist. |
trac user phsilva wrote on 2010-01-20 Thanks. You can close this issue if you want . I'll stay with distro installers for now and when/if I port my system to pip I'll report any issue I found. |
Original ticket http://projects.scipy.org/numpy/ticket/860 on 2008-07-24 by trac user phsilva, assigned to @dmcooke.
Numpy 1.1.0 works well with easy_install if called from command line, like:
easy_install numpy==1.1.0
BUT, if I declare numpy==1.1.0 as a dependency in my setup.py setup() function, then I got this error (actually I get the same errors with Numpy 1.0.4):
Traceback (most recent call last):
File "setup.py", line 116, in
platforms = chimera_platform)
File "/usr/lib/python2.5/distutils/core.py", line 151, in setup
dist.run_commands()
File "/usr/lib/python2.5/distutils/dist.py", line 974, in run_commands
self.run_command(cmd)
File "/usr/lib/python2.5/distutils/dist.py", line 994, in run_command
cmd_obj.run()
File "/usr/lib/python2.5/site-packages/setuptools-0.6c8-py2.5.egg/setuptools/command/develop.py", line 27, in run
File "/usr/lib/python2.5/site-packages/setuptools-0.6c8-py2.5.egg/setuptools/command/develop.py", line 102, in install_for_development
File "/usr/lib/python2.5/site-packages/easy_install.py", line 519, in process_distribution
File "/usr/lib/python2.5/site-packages/pkg_resources.py", line 522, in resolve
dist = best[req.key] = env.best_match(req, self, installer)
File "/usr/lib/python2.5/site-packages/pkg_resources.py", line 758, in best_match
return self.obtain(req, installer) # try and download/install
File "/usr/lib/python2.5/site-packages/pkg_resources.py", line 770, in obtain
return installer(requirement)
File "/usr/lib/python2.5/site-packages/easy_install.py", line 446, in easy_install
File "/usr/lib/python2.5/site-packages/easy_install.py", line 476, in install_item
File "/usr/lib/python2.5/site-packages/easy_install.py", line 655, in install_eggs
File "/usr/lib/python2.5/site-packages/easy_install.py", line 930, in build_and_install
File "/usr/lib/python2.5/site-packages/easy_install.py", line 919, in run_setup
File "/usr/lib/python2.5/site-packages/setuptools-0.6c8-py2.5.egg/setuptools/sandbox.py", line 27, in run_setup
File "/usr/lib/python2.5/site-packages/setuptools-0.6c8-py2.5.egg/setuptools/sandbox.py", line 63, in run
File "/usr/lib/python2.5/site-packages/setuptools-0.6c8-py2.5.egg/setuptools/sandbox.py", line 29, in
File "setup.py", line 96, in
"numpy == 1.1.0",
File "setup.py", line 89, in setup_package
"Elixir >= 0.5.2",
File "/tmp/easy_install-kOtWpP/numpy-1.1.0/numpy/distutils/core.py", line 150, in setup
File "setup.py", line 56, in configuration
# setup
File "/tmp/easy_install-kOtWpP/numpy-1.1.0/numpy/distutils/misc_util.py", line 839, in add_subpackage
File "/tmp/easy_install-kOtWpP/numpy-1.1.0/numpy/distutils/misc_util.py", line 822, in get_subpackage
File "/tmp/easy_install-kOtWpP/numpy-1.1.0/numpy/distutils/misc_util.py", line 769, in _get_configuration_from_setup_py
File "/home/henrique/work/chimera/chimera/setup.py", line 9, in configuration
# as published by the Free Software Foundation; either version 2
File "/tmp/easy_install-kOtWpP/numpy-1.1.0/numpy/distutils/misc_util.py", line 839, in add_subpackage
File "/tmp/easy_install-kOtWpP/numpy-1.1.0/numpy/distutils/misc_util.py", line 822, in get_subpackage
File "/tmp/easy_install-kOtWpP/numpy-1.1.0/numpy/distutils/misc_util.py", line 769, in _get_configuration_from_setup_py
File "/home/henrique/work/chimera/chimera/setup.py", line 223, in configuration
File "/tmp/easy_install-kOtWpP/numpy-1.1.0/numpy/distutils/misc_util.py", line 1039, in add_include_dirs
AttributeError: 'NoneType' object has no attribute 'extend'
This seems similar to issue #679. Numpy distutils was expecting something which setuptools distutils (only in dependency mode) overrides and throw away.
Follows a patch/hack against 1.1.0 that worked to me (for 1.0.4 it's just the same hack on different chunck).
What the plan for distutils based installers (like setuptools) when numscons became the default? I'm not going to integrate my package in distribution yet, so I'm using easy_install which seems to be the easiest solution available. Any other idea about how to distribute Numpy? I was thinking about try to create an egg, but then I'll need to have one for each different setup (distro/arch) that I would like to use, right?
Thanks a lot, and congrats for the great work with Numpy!
The text was updated successfully, but these errors were encountered: