Re: ImportError: No module named multiarray (is back)
Hi, Le 22/11/2011 20:59, Zbigniew Jędrzejewski-Szmek a écrit : > On 11/22/2011 04:38 PM, Mathieu Malaterre wrote: >> [CC me please] >> [...] > This is a very unfortunate interaction between Python semantics (prepend > realpath of script to sys.path) and the Debian way of installing modules > (symlink farm between Python versions). Nevertheless, I think that > Python is wrong, and Debian is right. What Python does means that it is > not possible to run a Python script from /tmp or any other directory > writeable by other users securely. Even if this is by design, it is not > advertised well (or maybe even at all?) and is thus a big security hole. Could you request a doc edition on bugs.python.org? Thanks. You’ll likely get feedback from Nick Coghlan, one of the core developers most knowledgeable about the import system. > You can wait for Python 3.2 and the PEP 3149 (ABI version tagged .so > files) which makes the Debian-style symlinks farms unnecessary (just > joking :). Are you really joking? I thought that PEP 3147 and PEP 3149 were especially designed to help Debian and Ubuntu get rid of their symlinks farms. Regards -- To UNSUBSCRIBE, email to debian-python-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/4ecd1d31.7080...@netwok.org
Re: ImportError: No module named multiarray (is back)
[CC me please] hi all, 2011/11/22 Zbigniew Jędrzejewski-Szmek : [...] > Hi Mathieu, > > I don't think that your problem can be solved in any simple way, apart from > physically moving the script or providing a second helper script. > > This is a very unfortunate interaction between Python semantics (prepend > realpath of script to sys.path) and the Debian way of installing modules > (symlink farm between Python versions). Nevertheless, I think that Python is > wrong, and Debian is right. What Python does means that it is not possible > to run a Python script from /tmp or any other directory writeable by other > users securely. Even if this is by design, it is not advertised well (or > maybe even at all?) and is thus a big security hole. Just like everything in computer, just add another extra layer of indirection. However this is exaclty that layer of indirection that I need help to get working. [Copy pasted from http://lists.debian.org/debian-python/2011/11/msg00059.html] Jakub Wilk : > It'd tad easier for us to help you if you explained what you were trying to > do. Short answer: trying to package the cellprofiler application. That is all. I can get the application working from my $HOME or /tmp no problem. However as soon as I move the CellProfiler.py within some know python location things behave differently. Package is at: http://mentors.debian.net/package/cellprofiler However it wont be very usefull as it wont work from the installed path. Thanks -- Mathieu -- To UNSUBSCRIBE, email to debian-python-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/CA+7wUsz5ioRp153BJv5DGKz_o=fPxXH+QV=zg9mvsgjecjy...@mail.gmail.com
Re: ImportError: No module named multiarray (is back)
* Mathieu Malaterre , 2011-11-22, 16:38: $ python /usr/lib/python2.7/dist-packages/CellProfiler.py Traceback (most recent call last): File "/usr/lib/python2.7/dist-packages/CellProfiler.py", line 17, in import numpy as np File "/usr/share/pyshared/numpy/__init__.py", line 136, in import add_newdocs File "/usr/share/pyshared/numpy/add_newdocs.py", line 9, in from numpy.lib import add_newdoc File "/usr/share/pyshared/numpy/lib/__init__.py", line 4, in from type_check import * File "/usr/share/pyshared/numpy/lib/type_check.py", line 8, in import numpy.core.numeric as _nx File "/usr/share/pyshared/numpy/core/__init__.py", line 5, in import multiarray ImportError: No module named multiarray [...] It'd tad easier for us to help you if you explained what you were trying to do. But how do I load this file ? What do you mean by "load"? -- Jakub Wilk -- To UNSUBSCRIBE, email to debian-python-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/2022200721.ga9...@jwilk.net
Re: ImportError: No module named multiarray (is back)
On 11/22/2011 04:38 PM, Mathieu Malaterre wrote: [CC me please] Hi all, The issue with multiarray came back. I manage to get things working for the tifffile package installing the tifffile.py into /usr/bin/tiffile (removing the py extension). However another package is now failing: CellProfiler. Symptoms are: $ python /usr/lib/python2.7/dist-packages/CellProfiler.py Traceback (most recent call last): File "/usr/lib/python2.7/dist-packages/CellProfiler.py", line 17, in import numpy as np File "/usr/share/pyshared/numpy/__init__.py", line 136, in import add_newdocs File "/usr/share/pyshared/numpy/add_newdocs.py", line 9, in from numpy.lib import add_newdoc File "/usr/share/pyshared/numpy/lib/__init__.py", line 4, in from type_check import * File "/usr/share/pyshared/numpy/lib/type_check.py", line 8, in import numpy.core.numeric as _nx File "/usr/share/pyshared/numpy/core/__init__.py", line 5, in import multiarray ImportError: No module named multiarray Hi Mathieu, I don't think that your problem can be solved in any simple way, apart from physically moving the script or providing a second helper script. This is a very unfortunate interaction between Python semantics (prepend realpath of script to sys.path) and the Debian way of installing modules (symlink farm between Python versions). Nevertheless, I think that Python is wrong, and Debian is right. What Python does means that it is not possible to run a Python script from /tmp or any other directory writeable by other users securely. Even if this is by design, it is not advertised well (or maybe even at all?) and is thus a big security hole. Now if I do: $ cp /usr/lib/python2.7/dist-packages/CellProfiler.py /tmp $ python /tmp/CellProfiler.py Traceback (most recent call last): File "/tmp/CellProfiler.py", line 193, in os.chdir(os.path.join(root, script_path)) OSError: [Errno 2] No such file or directory: '/tmp/cellprofiler/cpmath' Could someone please point me to some documentation get rid definitely of this kind of issues. You can wait for Python 3.2 and the PEP 3149 (ABI version tagged .so files) which makes the Debian-style symlinks farms unnecessary (just joking :). Best, Zbyszek > I'd like to keep as much as the original package as possible, therefore I would like to keep the /usr/lib/python2.7/dist-packages/CellProfiler.py (or pyshared equivalent). But how do I load this file ? Thanks very much. -- To UNSUBSCRIBE, email to debian-python-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/4ecbff19.8080...@in.waw.pl
ImportError: No module named multiarray (is back)
[CC me please] Hi all, The issue with multiarray came back. I manage to get things working for the tifffile package installing the tifffile.py into /usr/bin/tiffile (removing the py extension). However another package is now failing: CellProfiler. Symptoms are: $ python /usr/lib/python2.7/dist-packages/CellProfiler.py Traceback (most recent call last): File "/usr/lib/python2.7/dist-packages/CellProfiler.py", line 17, in import numpy as np File "/usr/share/pyshared/numpy/__init__.py", line 136, in import add_newdocs File "/usr/share/pyshared/numpy/add_newdocs.py", line 9, in from numpy.lib import add_newdoc File "/usr/share/pyshared/numpy/lib/__init__.py", line 4, in from type_check import * File "/usr/share/pyshared/numpy/lib/type_check.py", line 8, in import numpy.core.numeric as _nx File "/usr/share/pyshared/numpy/core/__init__.py", line 5, in import multiarray ImportError: No module named multiarray Now if I do: $ cp /usr/lib/python2.7/dist-packages/CellProfiler.py /tmp $ python /tmp/CellProfiler.py Traceback (most recent call last): File "/tmp/CellProfiler.py", line 193, in os.chdir(os.path.join(root, script_path)) OSError: [Errno 2] No such file or directory: '/tmp/cellprofiler/cpmath' Could someone please point me to some documentation get rid definitely of this kind of issues. I'd like to keep as much as the original package as possible, therefore I would like to keep the /usr/lib/python2.7/dist-packages/CellProfiler.py (or pyshared equivalent). But how do I load this file ? Thanks very much. -- To UNSUBSCRIBE, email to debian-python-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/CA+7wUswQVDddJJiXaJFk8gNmYMW9L4W_3GtSWKDV9LAL=sa...@mail.gmail.com
