[Numpy-discussion] take not respecting masked arrays?

Pierre GM pgmdevlist@gmail....
Mon Mar 1 00:40:08 CST 2010


On Feb 28, 2010, at 11:12 PM, Charles R Harris wrote:
> 
> 
> ______
> 
> Ah, Pierre, now that you are here... ;) Can you take a look at the invalid value warnings in the masked array tests and maybe fix them up by turning  off the warnings where appropriate? I'd do it myself except that I hesitate to touch masked array stuff.

Chuck, did you just hijack the thread ;) ?
To replace thiings in context: a few weeks ago, we had a global flag in numpy.ma that prevented warnings to be printed. Now, the warnings are handled in a case-by-case basis in the numy.ma functions. The problem is that when a numpy function is called on a masked array instead of its numpy.ma equivalent, the warnings are not trapped (that's what happen in the tests). In order to trap them, I'll have to use a new approach (something like __array_prepare__), which is not necessarily difficult but not trivial either. I should have plenty of free time in the next weeks. I'm afraid that won't be on time for the 2.0 release, though, sorry.


More information about the NumPy-Discussion mailing list