…piled with MSVC. Allows different compiler to be used than build platform default.
…ible segfaults in specfun interface due to n=0 or n=1.
…t finite numbers. Changed behavior of all and any to work over raveled arrays (Numeric's alltrue and sometrue still give old behavior)
…es made by Gary Pajer
…ults <float> and not 0-rank array, additional asarray is applied in sinc.
…changed if tests are failing.
…meters are ignored at the end of the interval.
…p a few of the interfaces to specfun. All of the specfun.f interfaces need to be checked to make sure that a large enough array is passed in. Some of the specfun routines assume that the array is at least size 2 (hard coded references to array(0) and array(1) for example). Unless the Python interface checks to make sure that it is being called with a large enough value, the array passed to the fortran routine can be too small. This has caused segfaults in the past and it is not entirely cleaned up.
…ach other and Abrahmowitz and Stegan. Started fixing up other tests in special that fail.
…outines when nans in matrix
…est failures in special.
…ed logm for certain complex matrices with zero imaginary part.
…w scipy sdist's and installs ok (again) from tar-balls.
…ices. Added better docs for special.errprint. Changed default of errprint to 0
…efined and evaluated at 0.0