Re: [Numpy-discussion] Remove numpy/compat/_inspect.py ?
The docstring at the beginning of the module is still relevant AFAIK: it was about decreasing import times. See http://mail.scipy.org/pipermail/numpy-discussion/2009-October/045981.html On Fri, Aug 1, 2014 at 10:27 AM, Charles R Harris wrote: > Hi All, > > The _inspect.py function looks like a numpy version of the python inspect > function. ISTR that is was a work around for problems with the early python > versions, but that would have been back in 2009. > > Thoughts? > > Chuck > > ___ > NumPy-Discussion mailing list > NumPy-Discussion@scipy.org > http://mail.scipy.org/mailman/listinfo/numpy-discussion > > ___ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion
[Numpy-discussion] Remove numpy/compat/_inspect.py ?
Hi All, The _inspect.py function looks like a numpy version of the python inspect function. ISTR that is was a work around for problems with the early python versions, but that would have been back in 2009. Thoughts? Chuck ___ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion
Re: [Numpy-discussion] OSX wheels for older numpy versions on pypi
On Thu, Jul 31, 2014 at 1:40 PM, Matthew Brett wrote: > On Thu, Jul 31, 2014 at 4:56 AM, Olivier Grisel > wrote: >> 2014-07-31 0:52 GMT+02:00 Matthew Brett : >>> Hi, >>> >>> I took the liberty of uploading OSX wheels for some older numpy >>> versions to pypi. These can be useful for testing, or when building >>> your own wheels to be compatible with earlier numpy versions - see: >>> >>> http://stackoverflow.com/questions/17709641/valueerror-numpy-dtype-has-the-wrong-size-try-recompiling/18369312#18369312 >>> >>> There are currently wheels for >>> >>> numpy 1.5.1 py27 >>> numpy 1.6.0 py27 >>> numpy 1.6.1 py27 >>> numpy 1.7.1 py27, 32, 33, 34 >>> >>> These are all compiled against ATLAS: >>> >>> https://github.com/matthew-brett/numpy-atlas-binaries >>> >>> install with e.g. >>> >>> pip install numpy==1.6.1 >> >> Thanks, this is very helpful for project maintainers who have to >> switch between versions to reproduce bugs reported by users. >> >> Do you plan do do the same for scipy? As scipy is even slower to build >> that would be even more helpful. > > Sure, I built and uploaded: > > scipy-0.12.0 py27 > scipy-0.13.0 py27, 33, 34 I uploaded 0.11.0 and 0.10.0 for py27 in the meantime, Cheers, Matthew ___ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion
[Numpy-discussion] working with numpy object arrays
In the example code below, is it possible to return an array of all the ".a" values of the MyClass objects as stored in the object array "a"? I am successfully able to retrieve the "a" attributes if I loop through the array elements one by one, but I cannot do a whole-array operation to retrieve the "a" attributes. Is there any way to retrieve all the "a" attributes of the MyClass objects all at once, or do I have to loop through all elements of "array" one-by-one? Thanks for any help, Catherine import numpy class MyClass(object): def __init__(self, a): self.a = a def add(self, b, c): self.a += b+c def return_a(self): return self.a array = numpy.empty((2,2), dtype=object) for i in xrange(0, 2): for j in xrange(0, 2): array[i,j] = MyClass(i+j) for i in xrange(0, 2): for j in xrange(0, 2): array[i,j].add(i, j) print "(%i,%i) = %i" % (i, j, array[i,j].a) try: array_a = array[:,:].a print "a values =",array_a except AttributeError: print "Unable to access a attributes of array as a whole." array_a = numpy.empty((2,2)) for i in xrange(0, 2): for j in xrange(0, 2): array_a[i,j] = array[i,j].a print "a values =",array_a ___ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion
Re: [Numpy-discussion] OSX wheels for older numpy versions on pypi
On Thu, Jul 31, 2014 at 4:56 AM, Olivier Grisel wrote: > 2014-07-31 0:52 GMT+02:00 Matthew Brett : >> Hi, >> >> I took the liberty of uploading OSX wheels for some older numpy >> versions to pypi. These can be useful for testing, or when building >> your own wheels to be compatible with earlier numpy versions - see: >> >> http://stackoverflow.com/questions/17709641/valueerror-numpy-dtype-has-the-wrong-size-try-recompiling/18369312#18369312 >> >> There are currently wheels for >> >> numpy 1.5.1 py27 >> numpy 1.6.0 py27 >> numpy 1.6.1 py27 >> numpy 1.7.1 py27, 32, 33, 34 >> >> These are all compiled against ATLAS: >> >> https://github.com/matthew-brett/numpy-atlas-binaries >> >> install with e.g. >> >> pip install numpy==1.6.1 > > Thanks, this is very helpful for project maintainers who have to > switch between versions to reproduce bugs reported by users. > > Do you plan do do the same for scipy? As scipy is even slower to build > that would be even more helpful. Sure, I built and uploaded: scipy-0.12.0 py27 scipy-0.13.0 py27, 33, 34 Are there any others you need? Cheers, Matthew ___ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion
Re: [Numpy-discussion] ANN: NumPy 1.9.0 beta 2 release
On 31.07.2014 05:20, Charles R Harris wrote: > > > I just noticed that we are getting a couple of warnings on some platforms. > ... > > *USE_CHROOT=1 ARCH=i386 DIST=trusty PYTHON=3.4* > > /usr/local/lib/python3.4/dist-packages/numpy/distutils/cpuinfo.py:120: > UserWarning: [Errno 2] No such file or directory: '/proc/cpuinfo' > > warnings.warn(str(e), UserWarning) > > Not sure about the second one. > this should harmless, the chroot we use to test 32 bit here does not have the proc filesystem mounted, we could mount it but this distutils feature should not be relevant for travis. ___ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion
Re: [Numpy-discussion] OSX wheels for older numpy versions on pypi
2014-07-31 0:52 GMT+02:00 Matthew Brett : > Hi, > > I took the liberty of uploading OSX wheels for some older numpy > versions to pypi. These can be useful for testing, or when building > your own wheels to be compatible with earlier numpy versions - see: > > http://stackoverflow.com/questions/17709641/valueerror-numpy-dtype-has-the-wrong-size-try-recompiling/18369312#18369312 > > There are currently wheels for > > numpy 1.5.1 py27 > numpy 1.6.0 py27 > numpy 1.6.1 py27 > numpy 1.7.1 py27, 32, 33, 34 > > These are all compiled against ATLAS: > > https://github.com/matthew-brett/numpy-atlas-binaries > > install with e.g. > > pip install numpy==1.6.1 Thanks, this is very helpful for project maintainers who have to switch between versions to reproduce bugs reported by users. Do you plan do do the same for scipy? As scipy is even slower to build that would be even more helpful. -- Olivier http://twitter.com/ogrisel - http://github.com/ogrisel ___ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion