scipy.weave.inline compilation fails with gcc 4.3

[Sameer Morar]

Binary package hint: python-scipy

The compiler fails with the following error:

<weave: compiling>
In file included from /usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/applics.h:400,
                 from /usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/vecexpr.h:32,
                 from /usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/vecpick.cc:16,
                 from /usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/vecpick.h:293,
                 from /usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/vector.h:449,
                 from /usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/tinyvec.h:430,
                 from /usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/array-impl.h:44,
                 from /usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/array.h:32,
                 from /home/smorar/.python25_compiled/sc_b1d4a37f384bb7ae593fe0b86178e5990.cpp:11:
/usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/mathfunc.h: In static member function ‘static long int blitz::_bz_abs<long int>::apply(long int)’:
/usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/mathfunc.h:45: error: ‘labs’ is not a member of ‘std’
In file included from /usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/array/funcs.h:29,
                 from /usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/array/newet.h:29,
                 from /usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/array/et.h:27,
                 from /usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/array-impl.h:2515,
                 from /usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/array.h:32,
                 from /home/smorar/.python25_compiled/sc_b1d4a37f384bb7ae593fe0b86178e5990.cpp:11:
/usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/funcs.h: In static member function ‘static int blitz::Fn_abs<int>::apply(int)’:
/usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/funcs.h:509: error: call of overloaded ‘abs(int&)’ is ambiguous
/usr/include/c++/4.3/cmath:99: note: candidates are: double std::abs(double)
/usr/include/c++/4.3/cmath:103: note: float std::abs(float)
/usr/include/c++/4.3/cmath:107: note: long double std::abs(long double)
/usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/funcs.h: In static member function ‘static long int blitz::Fn_abs<long int>::apply(long int)’:
/usr/lib/python2.5/site-packages/scipy/weave/blitz/blitz/funcs.h:530: error: ‘labs’ is not a member of ‘std’

This problem can be worked around by changing /usr/bin/g++ from g++-4.3 to point to g++-4.2

I am using Intrepid
root@spider:/usr/bin# dpkg -l|grep python-scipy
ii python-scipy 0.6.0-11 scientific tools for Python

[Sameer Morar (smorar)]

Attached is some python code which demonstrates the bug. To reproduce, please follow the following steps:
1) Ensure that gcc-4.3 is installed:

root@spider:/usr/bin# ls -la g++
lrwxrwxrwx 1 root root 7 2008-11-27 02:37 g++ -> g++-4.3

2) Remove pre-compiled python snippets

smorar@spider:~$ rm -rf /home/smorar/.python25_compiled/

3) Run code

smorar@spider:~$ python weave_test.py
The compilation will fail with the above error.

To demonstrate that this is related to the latest gcc:
1) Ensure that gcc-4.2 is installed

root@spider:/usr/bin# rm g++
root@spider:/usr/bin# ln -s g++-4.2 g++

2) Remove pre-compiled python snippets

smorar@spider:~$ rm -rf /home/smorar/.python25_compiled/

3) Run code

smorar@spider:~$ python weave_test.py
The compilation will succeed (with many compiler warnings), and the python / c code produces the expected output.

[Matthias Klose (doko)]

an include of cstdlib seems to be missing.

[Benjamin Horstman (yodenss)]

I can confirm the bug and workaround of linking g++-4.2.

For reference, here is the upstream ticket:
http://www.scipy.org/scipy/scipy/ticket/739

What Matthias has noted is present in the debian patch.

And the original (as far as I can tell) mention of this problem on the debian list, more than a year ago:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=455661

[Luca Falavigna (dktrkranz)]

Fixed upstream with new 0.7.0.

[Launchpad Janitor (janitor)]

This bug was fixed in the package python-scipy - 0.7.0-0ubuntu1

---------------
python-scipy (0.7.0-0ubuntu1) jaunty; urgency=low

  * New upstream version (LP: #285936, #299375, #302649).
  * Apply changes from Debian's python-modules team, thanks!
  * debian/rules: make python-scipy ready for Python 2.6 (LP: #336239):
    - Include /usr/share/python/python.mk.
    - Pass $(py_setup_install_args) macro to setup.py install.
    - Remove some files present in new dist-packages directory too.

 -- Luca Falavigna <email address hidden> Mon, 02 Mar 2009 17:28:08 +0100

[Benjamin Horstman (yodenss)]

Note that according to the scipy bug tracker, this isn't going to be fixed until scipy 0.7.1, which means that this bug still exists in ubuntu 9.04.

[Sameer Morar (smorar)]

This bug also exists in ubuntu 9.10:

smorar@spud:~/$ dpkg -l|grep scipy
ii python-scipy 0.7.0-2

[Sameer Morar (smorar)]

I have changed the status in python-scipy (ubuntu) to incomplete, as it should only be tagged as fixed when scipy 0.7.1 is available in the ubuntu repository.

[Jakob Runge (jakobrunge)]

The weave bug seems to be still present in scipy 0.7.1

I am running Ubuntu 9.10 (Karmic) and have scipy installed from source (since it is not in the repos) without problems, the required Atlas, ... libraries were installed using synaptic

ipython:
In [2]: scipy.__version__
Out[2]: '0.7.1'

Running a weave.blitz test routine:
"...
weave.inline(code, vars, type_converters = weave.converters.blitz)
..."

Output is:
starting weavefunc..
In file included from /usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/applics.h:394,
                 from /usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/vecexpr.h:26,
                 from /usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/vecpick.cc:16,
                 from /usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/vecpick.h:287,
                 from /usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/vector.h:443,
                 from /usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/tinyvec.h:424,
                 from /usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/array-impl.h:38,
                 from /usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/array.h:26,
                 from /home/jakob/.python26_compiled/sc_6211de8cbbdd4538661eae1a48c2a9890.cpp:11:
/usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/mathfunc.h: In static member function ‘static long int blitz::_bz_abs<long int>::apply(long int)’:
/usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/mathfunc.h:45: error: ‘labs’ is not a member of ‘std’
In file included from /usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/array/funcs.h:23,
                 from /usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/array/newet.h:23,
                 from /usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/array/et.h:21,
                 from /usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/array-impl.h:2509,
                 from /usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/array.h:26,
                 from /home/jakob/.python26_compiled/sc_6211de8cbbdd4538661eae1a48c2a9890.cpp:11:
/usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/funcs.h: In static member function ‘static int blitz::Fn_abs<int>::apply(int)’:
/usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/funcs.h:503: error: call of overloaded ‘abs(int&)’ is ambiguous
/usr/include/c++/4.4/cmath:94: note: candidates are: double std::abs(double)
/usr/include/c++/4.4/cmath:98: note: float std::abs(float)
/usr/include/c++/4.4/cmath:102: note: long double std::abs(long double)
/usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/funcs.h: In static member function ‘static long int blitz::Fn_abs<long int>::apply(long int)’:
/usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/funcs.h:524: error: ‘labs’ is not a member of ‘std’
In file included from /usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/applics.h:394,
                 from /usr/local/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/vecexpr.h:26...

[Jakob Runge (jakobrunge)]

changing the compiler to g++-4.2 it works... Karmic comes with g++-4.4 though

[Sameer Morar (smorar)]

This bug still exists in ubuntu 10.10:

smorar@spud:~/$ dpkg -l|grep scipy
ii python-scipy 0.7.2-2

[Matthias Klose (doko)]

Sameer, Stefano, what is the problem in maverick?

[Stefano Rivera (stefanor)]

Matthias: Haven't had a proper look at these headers, but the test case attached to this bug seems to show up some bugs in them.

Full output attached

In file included from /usr/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/applics.h:394,
                 from /usr/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/vecexpr.h:26,
                 from /usr/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/vecpick.cc:16,
                 from /usr/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/vecpick.h:287,
                 from /usr/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/vector.h:443,
                 from /usr/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/tinyvec.h:424,
                 from /usr/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/array-impl.h:38,
                 from /usr/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/array.h:26,
                 from /home/stefanor/.python26_compiled/sc_3ec397f0c53c3b3e50e2375deba7b7ed1.cpp:11:
/usr/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/mathfunc.h: In static member function ‘static long int blitz::_bz_abs<long int>::apply(long int)’:
/usr/lib/python2.6/dist-packages/scipy/weave/blitz/blitz/mathfunc.h:45: error: ‘labs’ is not a member of ‘std’

[Stefano Rivera (stefanor)]

Err, meant to add, obviously this is now an issue with 4.4

[Sameer Morar (smorar)]

The fix for this bug is simple. As per the debian patch linked to below, one needs to add an include for cstdlib above the BZ_NAMESPACE(blitz) call in the following files:
/usr/share/pyshared/scipy/weave/blitz/blitz/blitz.h
/usr/share/pyshared/scipy/weave/blitz/blitz/mathfunc.h

http://bugs.debian.org/cgi-bin/bugreport.cgi?msg=10;filename=blitz%2B%2B.patch;att=1;bug=455661

This fix works for me by manually editing those files.

[Sameer Morar (smorar)]

Patch to fix this bug

[Sameer Morar (smorar)]

Updated patch - added an cstdlib include to prettyprint.h
Change reference to launchpad bug number.
Updated the Patch Tags to only use a description and bug-ubuntu fields.

[Stefano Rivera (stefanor)]

Thanks for the patch, I'll forward this to Debian

[Sameer Morar (smorar)]

Here is a patch for lucid

[Sameer Morar (smorar)]

Here is a patch for lucid

[Launchpad Janitor (janitor)]

This bug was fixed in the package python-scipy - 0.7.2-2ubuntu1

---------------
python-scipy (0.7.2-2ubuntu1) maverick; urgency=low

  * debian/patches/blitz++.patch:
    - Fix scipy.weave.inline compilations (LP: #302649)
 -- Sameer Morar <email address hidden> Wed, 29 Sep 2010 13:42:10 +0200

[Stefano Rivera (stefanor)]

Uploaded to lucid-proposed, pending SRU team verification.

[Martin Pitt (pitti)]

Accepted python-scipy into lucid-proposed, the package will build now and be available in a few hours. Please test and give feedback here. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation how to enable and use -proposed. Thank you in advance!

[Sameer Morar (smorar)]

The new package works for me.

[Sameer Morar (smorar)]

Addendum to comment 24:

To verify the fix, python-dev needs to be installed. Will open this as a separate bug, as python-scipy does not recommend or suggest python-dev to be installed.

[Launchpad Janitor (janitor)]

This bug was fixed in the package python-scipy - 0.7.0-2ubuntu0.1

---------------
python-scipy (0.7.0-2ubuntu0.1) lucid-proposed; urgency=low

  * Fix scipy.weave.inline compilations (LP: #302649)
 -- Sameer Morar <email address hidden> Thu, 30 Sep 2010 15:32:13 +0200