search

SNAPlabgroup / ANLffr

A set of tools to analyze and interpret auditory steady-state responses, particularly the subcortical kind commonly known as frequency-following responses (FFRs).
BSD 3-Clause "New" or "Revised" License
11 stars 18 forks source link

Backwards compatibility fixes #5

Closed ghost closed 10 years ago

ghost commented 10 years ago

See commit comments.

ghost commented 10 years ago

See the latest commit. nfft will default to the next greatest power of 2 than the number of time points if the user doesn't specify it. If what the user specifies is less than the number of time points, they should get a warning that it's going back to the default value (via logger.warn, but see if I implemented it correctly for the _get_nfft function)

ghost commented 10 years ago

OK, I'm noticing a warning now that I was not getting prior to the upstream merge I did.

/home/lennyv/virtualenvs/scipy_opt/local/lib/python2.7/site-packages/scipy/spatial/init.py:91: RuntimeWarning: numpy.dtype size changed, may indicate binary incompatibility from .qhull import *

scipy/spatial ?

haribharadwaj commented 10 years ago

For which import are you getting the warning?

ghost commented 10 years ago

"For which import are you getting the warning?"

It has to do with the dpss changeover; the warning goes away when i change:

from .dpss import dpss_windows

to 

from nitime.algorithms import dpss_windows

FWIW, I'm running numpy 1.8.0, scipy 0.13.3, nitime 0.4, all installed via pip.

haribharadwaj commented 10 years ago

@lennyvarghese .. Ok... So I merged this PR.... As far as the nfft issue goes, I thought just putting the _validate_parameters(.) call immediately after the _get_nfft(.) call would do it?

I'll look into the import issue for the new dpss.. doesn't ring a bell yet... I can work on the doc later today...

Thanks a lot!