Mailman 3 July 2008 - NumPy-Discussion

chararray __mod__ behavior
by Alan McIntyre July 19, 2008

July 19, 2008

This seems odd to me: >>> A=np.array([['%.3f','%d'],['%s','%r']]).view(np.chararray) >>> A % np.array([[1,2],[3,4]]) Traceback (most recent call last): File "<stdin>", line 1, in <module> File "/opt/local/lib/python2.5/site-packages/numpy/core/defchararray.py", line 126, in __mod__ newarr[:] = res ValueError: shape mismatch: objects cannot be broadcast to a single shape Is this expected behavior? The % gets broadcast as I'd expect for 1D arrays, but more dimensions fail as above. Changing the offending line in defchararray.py to "newarr.flat = res" makes it behave properly.

2 2

Another reference count leak: ticket #848
by Michael Abbott July 19, 2008

July 19, 2008

The attached patch fixes another reference count leak in the use of PyArray_DescrFromType. Could I ask that both this patch and my earlier one (ticket #843) be applied to subversion. Thank you. Definitely not enjoying this low level code. commit 80e1aca1725dd4cd8e091126cf515c39ac3a33ff Author: Michael Abbott <michael.abbott(a)diamond.ac.uk> Date: Tue Jul 8 10:10:59 2008 +0100 Another reference leak using PyArray_DescrFromType This change fixes two issues: a spurious ADDREF on a typecode returned from PyArray_DescrFromType and a return path with no DECREF. diff --git a/numpy/core/src/scalartypes.inc.src b/numpy/core/src/scalartypes.inc.src index 3feefc0..772cf94 100644 --- a/numpy/core/src/scalartypes.inc.src +++ b/numpy/core/src/scalartypes.inc.src @@ -1886,7 +1886,6 @@ static PyObject * if (!PyArg_ParseTuple(args, "|O", &obj)) return NULL; typecode = PyArray_DescrFromType(PyArray_@TYPE@); - Py_INCREF(typecode); if (obj == NULL) { #if @default@ == 0 char *mem; @@ -1904,7 +1903,10 @@ static PyObject * } arr = PyArray_FromAny(obj, typecode, 0, 0, FORCECAST, NULL); - if ((arr==NULL) || (PyArray_NDIM(arr) > 0)) return arr; + if ((arr==NULL) || (PyArray_NDIM(arr) > 0)) { + Py_XDECREF(typecode); + return arr; + } robj = PyArray_Return((PyArrayObject *)arr); finish:

5 12

f2py - how to use .pyf files?
by Fernando Perez July 19, 2008

July 19, 2008

Howdy, again f2py... Since I can't seem to figure out how to pass the --fcompiler option to f2py via setup.py/distutils, I decided to just do things for now via a plain makefile. But I'm struggling here too. The problem is this: the call f2py -c --fcompiler=gfortran -m text2 Text2.f90 works perfectly, and at some point in the output I see customize Gnu95FCompiler Found executable /usr/bin/gfortran and the result is a nice text2.so module. But I'd like to clean up a bit the python interface to the fortran routines, so I did the usual f2py -h text2.pyf Text2.f90 to create the .pyf, edited the pyf to adjust and 'pythonize' the interface, and then when I try to build using this pyf, I get a crash and the *same* gfortran option is now not recognized: maqroll[felipe_fortran]> f2py -c --fcompiler=gfortran text2.pyf Unknown vendor: "gfortran" Traceback (most recent call last): File "/usr/bin/f2py", line 26, in <module> main() File "/home/fperez/usr/opt/lib/python2.5/site-packages/numpy/f2py/f2py2e.py", line 560, in main run_compile() File "/home/fperez/usr/opt/lib/python2.5/site-packages/numpy/f2py/f2py2e.py", line 536, in run_compile ext = Extension(**ext_args) File "/home/fperez/usr/opt/lib/python2.5/site-packages/numpy/distutils/extension.py", line 45, in __init__ export_symbols) File "/usr/lib/python2.5/distutils/extension.py", line 106, in __init__ assert type(name) is StringType, "'name' must be a string" AssertionError: 'name' must be a string Note that it doesn't matter if I add Text2.f90 or not to the above call, it still fails. I could swear I'd done similar things in the past without any problems (albeit with f77 sources), and the user guide http://cens.ioc.ee/projects/f2py2e/usersguide/index.html#three-ways-to-wrap… gives instructions very much along the lines of what I'm doing. Are these changes since the integration into numpy, regressions, or mistakes on how I'm calling it? Thanks for any help, f

2 9

Re: [Numpy-discussion] [matplotlib-devel] Matplotlib and latest numpy
by Gael Varoquaux July 19, 2008

July 19, 2008

On Sat, Jul 19, 2008 at 01:25:51AM +0200, Gael Varoquaux wrote: > Am I wrong, or does matploib not build with current numpy svn? OK, Fernando told me that matplotlib builds fine with latest numpy on his box so Ienquired a bit more. The problem is that the build of matplotlib tries to include a header file that is generated automatically during the build of numpy (__multiarray_api.h). If you install numpy using "python setup.py install", this header file is inlcuded in the install. However I used "python setupegg.py develop" to install numpy. As a result the header file does not get put in the include directory. I guess this is thus a bug in the setupegg.py of numpy, but my knowledge of setuptools is way to low to be able to fix that. Cheers, Gaël

3 9

f2py_options in setup.py
by Fernando Perez July 18, 2008

July 18, 2008

Howdy, setup.py files with calls like f2py_options = ['--fcompiler=gfortran'], ... when building the extension object used to work. But now I'm trying the same thing with some f90 code and when running the build, it doesn't seem to go through to f2py: maqroll[felipe_fortran]> ./setup.py build running build running scons customize UnixCCompiler Found executable /usr/bin/gcc customize GnuFCompiler Found executable /usr/bin/g77 gnu: no Fortran 90 compiler found gnu: no Fortran 90 compiler found customize GnuFCompiler gnu: no Fortran 90 compiler found gnu: no Fortran 90 compiler found customize UnixCCompiler customize UnixCCompiler using scons Found executable /usr/bin/g++ etc... If I run f2py itself at the command line, I can give it the compiler flag correctly. But via setup.py it just doesn't seem to be working, and I'm at a loss as to what I'm doing wrong. Any help would be much appreciated. Cheers, f ps - is there an updated f2py guide on the scipy site somewhere? The guide I found was the old one from http://cens.ioc.ee/projects/f2py2e/ But I don't know if, after all the recent integration with numpy, all of it remains valid. It would be good to have f2py documented as part of numpy, since it's now part of it.

1 0

Re: [Numpy-discussion] 1.1.0 OSX Installer Fails Under 10.5.3?
by Christopher Burns July 18, 2008

July 18, 2008

I've been using bdist_mpkg to build the OSX Installer. I'd like to update the requirement documentation for the 1.1.1 release candidate to say "MacPython from python.org" instead of "System Python". bdist_mpkg specifies this, does anyone know how to override it? Chris On Tue, Jun 3, 2008 at 11:48 PM, J. Stark <j.stark(a)imperial.ac.uk> wrote: > Chris, > > many thanks. Could I suggest that this information be featured > prominently in the Read Me in the Installer, and perhaps also at > http://www.scipy.org/Download where this is given as the official > binary distribution for MacOSX. You might want to change the error > message too, since I think that some people will interpret "System > Python" to mean the default Python provided by the standard system > install. Since this is 2.5.1 on Leopard, the error message could be > confusing. > > <http://cirl.berkeley.edu/>

2 1

Type checking of arrays containing strings
by Arnar Flatberg July 18, 2008

July 18, 2008

Hi I need to check if my array (a) is of type `string`. That is, I dont know the number of characters beforehand, so I cant do a.dtype == '|S*' (* = (max) number of characters) Looking at my options, I see either a.dtype.kind == 'S' or a.dtype.type == np.string_, might be ok. Are these any of the preffered ways, or is there some other way? Thanks, Arnar

2 4

interleaved indexing
by Amir July 18, 2008

July 18, 2008

A very beginner question about indexing: let x be an array where n = len(x). I would like to create a view y of x such that: y[i] = x[i:i+m,...] for each i and a fixed m << n so I can do things like numpy.cov(y). With n large, allocating y is a problem for me. Currently, I either do for loops in cython or translate operations into correlate() but am hoping there is an easier way, maybe using fancy indexing or broadcasting. Memory usage is secondary to speed, though. Thanks.

2 1

Documtation updates for 1.1.1
by Charles R Harris July 17, 2008

July 17, 2008

Hi Stephan, I'm thinking it would nice to backport as many documentation updates to 1.1.1 as possible. It looks like the following steps should do the trick. 1) Make ptvirtan's changes for ufunc documentation. 2) Copy add_newdocs.py 3) Copy fromnumeric.py Does that look reasonable to you? Chuck

4 4

Numpy Trac malfunctions
by Pauli Virtanen July 17, 2008

July 17, 2008

Hi, Trac seems to malfunction again with permission problems. At http://projects.scipy.org/scipy/numpy/changeset/5447 there is Traceback (most recent call last): File "/usr/lib/python2.4/site-packages/trac/web/main.py", line 387, in dispatch_request dispatcher.dispatch(req) File "/usr/lib/python2.4/site-packages/trac/web/main.py", line 238, in dispatch resp = chosen_handler.process_request(req) File "/usr/lib/python2.4/site-packages/trac/versioncontrol/web_ui/changeset.py", line 188, in process_request prev = repos.get_node(new_path, new).get_previous() File "/usr/lib/python2.4/site-packages/trac/versioncontrol/cache.py", line 120, in get_node return self.repos.get_node(path, rev) File "/usr/lib/python2.4/site-packages/trac/versioncontrol/svn_fs.py", line 356, in get_node self.pool) File "/usr/lib/python2.4/site-packages/trac/versioncontrol/svn_fs.py", line 533, in __init__ self.root = fs.revision_root(fs_ptr, rev, self.pool()) SubversionException: ("Can't open file '/home/scipy/svn/numpy/db/revs/5447': Permission denied", 13) -- Pauli Virtanen

3 2